Fabrication and Field Emission Properties of Carbon Nanotubes

Transcription

1 Fabrication and Field Emission Properties of Carbon Nanotubes Peng Wang A dissertation submitted to the faculty of the University of North Carolina at Chapel Hill in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Physics and Astronomy. Chapel Hill 2007 Approved by: Professor Otto Zhou Professor Lu-Chang Qin Professor Jianping Lu Professor Hugon Karwowski Dr. Christopher Bower

2 2007 Peng Wang ALL RIGHTS RESERVED ii

3 ABSTRACT PENG WANG: Fabrication and Field Emission Properties of Carbon Nanotubes (Under the direction of Professor Otto Zhou) Research on the area of the fabrication of carbon nanotubes is fundamental and critical to the entire subject of carbon nanotubes. This dissertation describes an experiment to fabricate carbon nanotubes by the method of Microwave Plasma Enhanced Chemical Vapor Deposition (MPECVD) and the electron field emission properties of carbon nanotubes. A MPECVD system was built and used to fabricate the vertical aligned carbon nanotube film. Scanning electron microscope (SEM), Raman spectroscopy and transmission electron microscopy (TEM) were used to characterize the as-grown carbon nanotube samples. By using a metal-containing diblock copolymer catalyst, carbon nanotubes with a diameter of 4 to 7 nm were synthesized. The effect of growth parameters was studied and these parameters were optimized. The growth of high density (~ 10 9 /cm 2 ) and large coverage area (~ 1 cm 2 ) carbon nanotube film on glass substrate at low growth temperature was realized. Based on a series of experiments, the effects of oxygen atoms and Ti/N underlayer on the growth were studied. A series of experiments were evaluated to characterize the field emission properties of the various carbon nanotube cathodes. A simple technique of scratching the pattern surface by a cotton swab was found effective to activate more carbon nanotubes to emit. By using the techniques of photolithography and shadow mask, various carbon nanotube patterns were achieved in order to obtain high emission current density and a low threshold electric field. iii

4 The lowest threshold electric field was found to be 2.3 V/um. The highest current density was found to be 2.2 ma/cm 2 when the electric field was 4.7 V/um. Our work shows that it is feasible to provide greater control over the fabrication of carbon nanotubes so that more obstacles in the broad application of carbon nanotubes can be overcome. iv

5 ACKNOWLEDGEMENTS I would like to express my deep appreciation to Dr. Otto Zhou for providing me the mentoring and guiding in my PhD study on nano science. It has been a great life experience and a valuable fortunate to have you as a mentor and good friend. Your sensitivity to science and enthusiasm to work will keep inspiring me in the future. Special thanks to Dr. Jennifer Lu for providing the polymer catalyst. I have enjoyed all of our discussions and influenced a lot about by your attitude to science and work. Special thanks to Mr. Adam Hall for tutoring and helping in using SEM, and Dr. Yunyu Wang for helping in characterizing MPECVD. It was necessary and significant to my work. Special thanks to Dr. Shuxia Wang, a great partner to work with. Special thanks to machine shop staff (Mr. Philip Thompson, Mr. Neal E Johnson, Mr. Steve Medlin and Mr. Cliff Tysor) in Dept. of Physics & Astronomy for helping in building up the MPECVD system. I would also like to show my appreciation to David Bordelon, Jerry Zhang, Guang Yang, Lei An, Ramya Rajaram, Xiomara Calderon, Dr. Zejian Liu, Zhijun Liu, Christy Redmon, Yueh Lee, Dr. Changkun Dong, Dr Bo Gao, Dr. Yuan Cheng, Qi Qiu, Dr. Jane Geng, Dr. Qi Zhang, Zhongqiao Ren, Dmitri, Lloyd, Fred, Maggie Hudson, Beverley Metz, Barbara Szilvay and Celeste Winston. Without your help, it is impossible for this work to be done. To my parents: 我还要深深地感谢我的父母, 来自你们的无限的爱, 是我面对挑战和战胜困难的强大后盾 v

6 TABLE OF CONTENTS LIST OF FIGURES...x Chapter 1. Introduction 1.1. Definition of Carbon Nanotube (CNT) History of CNT Structure of CNT Properties of CNT Electronic Properties of CNT Optical Properties of CNT Mechanical Property of Carbon Nanotube Applications of CNT Motivation and Overview Synthesis of Carbon Nanotubes Introduction to Methods of CNT Synthesis Electric Arc Discharge Laser Ablation Thermal Chemical Vapor Deposition (CVD) Plasma Enhanced Chemical Vapor Deposition (PECVD) Experimental Setup for the MPECVD System Parameters Affecting the Growth of CNTs by MPECVD System vii

7 Selection of Catalyst Material Selection of Feeding Gases Effect of Plasma and Sheath Electric Field on CNT Growth in MPECVD Conclusion Fabrication of Small-diameter CNTs Using the Polymer Based Catalyst Introduction Catalyst Preparations Experimental Setup Catalyst Pretreatment Growth procedures Results Effect of Growth Temperature Effect of Growth Pressure Effect of Feedstock Gas Ratio Characterization Transmission Electron Microscopy (HRTEM) Vertically Alignment Conclusion Low Temperature Growth of CNTs Using Iron Film Catalyst Introduction Experiment Results and Discussion Conclusion..89 vii

8 5. Electron Field Emission of Carbon Nanotube Cathodes Introduction of Electron Field Emission Field Emission from Single Emitter and Multi-emitters Behavior of Carbon Nanotubes in Electric Field Techniques to Improve the Field Emission Properties of Carbon Nanotube Cathodes Experiment Field Emission and the effect of Cotton Swab Scratch Results Growth Pattern of Carbon Nanotubes on Glass Substrate and Field Emission Properties Introduction Synthesis and Characterization Field Emission Measurements and Results Summary Summary..131 Reference viii

9 LIST OF FIGURES Figure 1.1 One of the TEM images of MWNTs published in the Russian Journal of Physical Chemistry by Radushkevich and Lukyanovich in Figure of the relationships involving the key parameters that characterize the structure of an SWNT....5 r 1.3 Schematics to show the allowed K' s Some possible vectors specified by indices (n, m) to define the CNTs Calculated electronic 1D density of states per unit cell of a 2D graphene sheet for two (n, 0) zigzag nanotubes STM tunneling spectra of the isolated tube and comparison of the DOS obtained experimentally and calculation Schematic figure of the experimental setup for a field-effect transistor (FET) and transport characteristics of a FET Calculation of the energy separations E ii (d t ) for all (n, m) values as a function of nanotube diameter between 0.7 < d t < 3.0 nm Phonon dispersion relations of an armchair carbon nanotube, phonon density of states for the nanotube and comparison between nanotube and graphene The calculated Raman mode atomic displacements, frequencies and symmetries for selected normal modes for the (10, 10) nanotube modes Raman spectra of a SWNT sample Tensile loading of individual MWNTs Schematic drawing of the electric arc discharge experimental setup Schematic drawing of the laser ablation experimental setup Schematic drawing of the thermal CVD experimental setup Schematic drawing of the three growth models HRTEM picture of one example of middle growth. 31 x

10 2.6 Schematic drawing of the MPECVD system built to fabricate carbon nanotubes Image of the MPECVD system Schematic drawing of the details of the system s reaction chamber and boundary between the plasma and the substrate Schematic drawing of the sheath electric field SEM image and schematic drawing to show the growth direction of CNTs on various substrates SEM picture of showing the shape of the CNTs extending out from the vertical surface at the bottom of the object Distribution of the electric field and potential around the step-like protrusion on the substrate SEM image showing a square pattern with CNTs Schematic figure of the formation of the Fe ramp film on the edges of each pattern Molecular formula of the iron-containing diblock copolymer unit AFM image of the surface of the substrate after the spin-coating of the catalyst SEM pictures of the catalyst after the substrate has been treated by the O 2 plasma Effect of temperature on the growth of CNTs SEM pictures showing the effect of pressure on the growth of CNTs SEM pictures showing the CNTs grown at different gas ratios of C 2 H 2 /NH Raman spectrum of the samples with different gas ratios HRTEM images of sample number 7 and the diameter distribution of the as-grown carbon nanotubes SEM picture of the area scratched by the TEM grid SEM picture of the substrate after the acetone ultrasonic bath...67 xi

11 3.11 3Side view SEM picture of sample number SEM picture of the surface of sample number 2 after the scratch by the TEM grid The schematic drawing of the EPD method The schematic drawing of the screen printing method SEM images of the as-grown CNTs from different catalyst and growth conditions with the same growth temperature of 500 o C HRTEM image of the same sample as Fig SEM view of the cross section of the CNT film grown on glass substrate Digital camera image of the as-grown CNTs on glass substrate and SEM images of the as-grown CNTs on glass substrate SEM picture of the surface of the Fe catalyst after the annealing Schematic figures of the formation of Fe 2 O 3 on the surface of the substrate when the Fe film is annealed in air for either 550 o C or 700 o C SEM image of the growth result for the substrates with different treatments Crystal structure of Fe 2 O Schematic figure of the field emission process Various shapes of field emitters Electric field distribution around the tip and body of a whisker-like emitter Schematic drawing of the dielectrophoresis method SEM image of the final tungsten tip attached with carbon nanotube Schematic drawing of the cross section view of a Spindt-type cathode and SEM image of a single gated nanocone structure Proposed mechanism by which the individual CNTs assemble into uniform fibrils under an asymmetrical electric field SWNT-assembly and orientation in response to an applied DC electric field xii

12 5.9 Electrostatic attraction between two carbon nanotubes induced by a constant field across the electrodes Nanotube response to resonant alternating applied potentials In situ TEM observation of the electric-field-induced electron emission from carbon nanotubes SEM image of the shape of the nanotube before field emission and after field emission CNT was shortened after field emission Schematic drawing of the details of the pattern SEM images of the CNT pattern Digital camera image of the three instances of field emission The I-V curves of those three instances of field emission measurements Microscope image of the area after scratches Digital camera image of the substrate with the CNT pattern Top view SEM image of as-grown CNT film and the area scratched by tweezers Tilt angle SEM image of the CNT pattern HRTEM images of two types of carbon nanotubes Measured I-V curves of patterned CNT film and unpatterned CNT film Digital camera image of the emission from unpatterned CNT film and patterned CNT film Schematic drawing of the positional distribution of CNTs Top view of a 2-D CNT array Corresponding Fowler-Nordheim plots and linear fits to the Fowler-Nordheim plots from patterned and unpatterned films xiii

13 Chapter 1. Introduction 1.1 Definition of Carbon Nanotube (CNT) Carbon nanotube is one of the allotropes of carbon. The other allotropes of carbon are graphite, diamond, fullerene and amorphous carbon. The structure of a single-walled carbon nanotube (SWNT) can be regarded as by wrapping a one-atom-thick layer of graphite called graphene into a seamless tube-shaped cylinder. Multi-walled nanotubes (MWNT) consist of multiple layers of graphite rolled in on themselves to form a tube shape where the sheets of graphite are arranged in concentric cylinders. The ends of the tubes can be a hemisphere buckyball structure or open ended. The bonding between carbon atoms in the graphene is sp 2 bonding. The diameter of a typical SWNT is 0.8 nm to 2 nm. The diameter of MWNT can vary from a few nm to a hundred nm. The length of a nanotube or a bundle of nanotubes is in the range of several hundred nm to several millimeters. 1.2 History of CNT A large amount of academic and popular literature attributes the discovery of the carbon nanotube to Sumio Iijima of NEC in 1991 [1]. However, it is unclear who first discovered carbon nanotube. A 2006 editorial written by Marc Monthioux and Vladimir Kuznetsov in the journal Carbon has explored the original discovery of CNT and briefly described the development of CNT research prior to 1991 [2].

14 However, the history of discovery of SWNT is unambiguous. The formation of SWNT was first reported in the June 17 th issue of Nature in 1993 by two papers submitted independently by Iijima and Bethune, respectively [3, 4]. But the time of discovery of MWNT greatly predates 1991 and requires further discussion. The first mention of the possibility of forming carbon filaments from the thermal decomposition of gaseous hydrocarbon (methane) was reported in 1889 [5] in a patent that proposed the use of such filaments in the light bulbs that had just been presented by Edison at the Paris Universal Exposition the same year. But the observation of the sub-µm cavity and inner texture structure of MWNT is not possible without the transmission electron microscope (TEM), which became commercially available in The first TEM-based evidence for the tubular nature of some nano-sized carbon filaments formed by CO decomposition on iron substrate is believed to have been published by Radushkevich and Lukyanovich in 1952 in the Russian Journal of Physical Chemistry. But their discovery was unknown to most western scientists, who because of the Cold War lacked access to Russian scientific publications. According to [2], credit for the initial discovery of MWNT should be given to these two Russian scientists. Figure is one of the TEM images published in this paper to depict the structure of MWNT they had found. Many related reports followed during subsequent decades [6-10]. But none of these reports had as large an impact on the science community as the 1991 paper by Iijima. One reason is that the early research investigating carbon filaments and nanotubes was focused on building lightweight composite materials with superior mechanical properties to fulfill the needs of the space and aircraft industries. Fundamental physicists showed little interest in this field. More importantly, the materials, theory, tools and other aspects of science hadn t matured to where operating at the nano 2

15 scale would seem plausible to most scientists. Another reason is that the research on fullerenes worldwide earlier helped to promote the importance of carbon nanotubes. Figure 1.1 One of the TEM images of MWNTs published in the Russian Journal of Physical Chemistry by Radushkevich and Lukyanovich in

16 1.3 Structure of CNT As mentioned above, the structure of a SWNT can be regarded as a seamless cylinder derived from a single atomic layer of crystalline graphite sheet, which is known as a graphene. The diameter and chirality of an SWNT are uniquely characterized by the roll-up vector C h = na 1 + ma 2 (n, m) that connects crystallographically equivalent sites on a twodimensional (2D) graphene, where a 1 and a 2 are the graphene primitive lattice vectors, and n and m are integers. The translation vector, T, is directed along the SWNT axis and perpendicular to C h, which represents the circumferential direction of the SWNT; the magnitude of T corresponds to the length of the (n, m) SWNT unit cell. Figure shows the relationship between those vectors on a graphene sheet. The diameter of nanotube is given by: d t = C h / π = 3 1/2 a C-C (m 2 + mn + n 2 ) 1/2 /π where C h is the length of C h and a C-C is the C-C bond length (1.42 Ǻ). The chiral angle θ is given by: θ = tan -1 [3 1/2 m / (m + 2n)] Among a large number of possible C h vectors, there are two highly symmetric directions. They are termed zigzag and armchair and are designated by (n, 0) and (n, n), respectively. The chiral angles θ are 0 o and 30 o for zigzag and armchair, respectively. All nanotubes with other choices of C h are called chiral tubes without symmetry. High resolution transmission electron microscopy (HRTEM) and scanning tunneling microscopy (STM) are often used to explore the structure of SWNT. The diameter and values of n and m can be calculated by analyzing the diffraction pattern from the electron beam diffraction by HRTEM [11]. 4

17 C h Figure 1.2 Figure of the relationships involving the key parameters that characterize the structure of an SWNT [12]. 5

18 1.4 Properties of CNT Electronic Properties of CNT SWNT can be metals or semiconductors with different energy gaps that are sensitive to their diameter and helicity. The relationship between the electronic properties of carbon nanotubes and their structure can be understood within a band-folding picture. An isolated graphene is a zero-gap semiconductor. According to tight-binding calculation [13], the conduction band and valence band of graphite have linear dispersion and meet at the Fermi level at the K point in the Brillouin zone. The Fermi surface of an ideal graphene consists of six corner K points. When the piece of the graphene is rolling to form a tube, the periodic boundary conditions (C h k = 2πq) for the 1D carbon nanotubes of small diameter permit only a few wave vectors to exist in the circumferential direction, and these wave vectors K r satisfy the relation nλ = πd t, where λ = 2π/k is the de Broglie wavelength. The allowed set of r K' s, indicated by the lines in Figure 1.4.1, depends on the diameter and helicity of the tube, that is, on the indices (n, m). Whenever the allowed r K' s include the point K, the system is a metal with a nonzero density of states at the Fermi level. When the point K is not included, the system is a semiconductor with energy gaps of different sizes. Metallicity of Single-Walled Carbon Nanotube The general rules for the metallicity of the single-walled carbon nanotubes are as follows: armchair tubes (n, n) are metals; (n, m) tubes with n m = 3i, where i is a nonzero integer, are very tiny-gap semiconductors (but they show metallic properties at room temperature because the thermal energy overcomes the small band gaps); and all others are large-gap 6

19 semiconductors. Figure shows some possible vectors specified by indices (n, m) to define the CNTs and their metallicity. Density of State (DOS) Away from the K point, signature features in the density of states (DOS) of a material appear at the band edges, and are commonly referred to as van Hove singularities (VHS). These singularities are characteristic of the dimension of a system. In a one-dimensional system, the VHS appear as peaks. Therefore, SWNTs and other 1D materials are expected to exhibit spikes in the DOS due to the 1D nature of their band structure. The calculated onedimensional DOS per unit cell of a 2D graphite sheet and for two (n, 0) zigzag nanotubes are shown in Figure 1.4.3, from [13]. 7

20 r allowed K ' s K k y k x Figure 1.3 Schematics to show the allowed r K' s. Figure 1.4 Some possible vectors specified by indices (n, m) to define the CNTs. [14] 8

21 Figure 1.5 Calculated electronic 1D density of states per unit cell of a 2D graphene sheet for two (n, 0) zigzag nanotubes: (a) the (10, 0) nanotube, which has semiconducting behavior, and (b) the (9, 0) nanotube, which has metallic behavior. The dotted line shown in the figure is the density of states for the 2D graphene sheet. Metallicity of Multi-Walled Carbon Nanotube The energy differences E M 11(d t ) and E S 11(d t ) for metallic and semiconducting nanotubes between the highest-lying valence band singularity and the lowest-lying conduction band singularity in the 1D electronic DOS curves are expressed as follows: E M 11(d t ) = 6γ o a C-C /d t, and E S 11(d t ) = 2γ o a C-C /d t (1.4.1) 9

22 where a C-C is the C-C bond length (1.42 Ǻ) and γ o = 2.9 ev is the carbon-carbon transfer energy. The band gap for isolated semiconducting carbon nanotubes is proportional to the reciprocal nanotube diameter d t. At a nanotube diameter of d t ~ 3 nm, the bandgap becomes comparable to thermal energies at room temperature. A MWNT is composed of a set of coaxially arranged SWNTs of different radii. For each individual tube, the structure can be different and be either metallic or semiconducting. The metallicity of this MWNT is determined by the outmost shell. The diameter of the outmost tube of MWNTs is determined by the growth process. It is typically of the order of a few tens of nm. Therefore, most MWNT show the metallic character at room temperature. The Measurement of Indices (n, m) and DOS by STM and STS Experimentally, Scanning Tunneling Microscopy (STM) and Spectroscopy (STS) provide the precise means to explore the theoretical predictions about the electronic properties of carbon nanotubes, since these techniques are capable of resolving simultaneously the atomic structure and electronic DOS of a material. A high resolution image of a SWNT exhibits a graphite-like honeycomb lattice. The (n, m) indices were obtained from experimentally measured values of the chiral angle and diameter. The chiral angle was measured between the zigzag (n, 0) direction and the tube axis. The measurement of di/dv in the STS mode of a scanning tunneling microscope yields a signal that is proportional to the 1D density of states. This enables the relationship between the structure of carbon nanotubes and their electronic properties to be examined. Two sets of STM images and DOS figures are shown in Figure for metallic SWNT and semiconducting SWNT, respectively. 10

23 Figure 1.6 (a) STM tunneling spectra of the isolated tube. The inset shows an atomic resolution image of this tube. (b) STM image of a SWNT on the surface of a rope. (c) Comparison of the DOS obtained experimentally (upper curve) and a π-only tight-binding calculation for the (13, 7) SWNT (second curve from top). The calculated DOS for a (12, 6) tube is included for comparison. (d) Comparison of the DOS obtained experimentally (upper curve) and calculation for the (10, 0) SWNT (lower curve). The image and figure are from [15, 16]. Example of a FET Setup using Carbon Nanotubes The unique electronic property and nm size of carbon nanotubes bolster the idea of nanoscale electronic devices in future. The electrical characterization of individual SWNT 11

24 molecules has been realized by both bottom-up chemical synthesis of these materials and top-down lithographic techniques for making electrical contacts. A room temperature field effect transistor (FET) with semiconducting nanotube was made in 1998 [17]. Figure (a) shows the schematics of the experimental setup. Figure (b) depicts the currentvoltage (I-V) characteristics for semiconducting nanotubes. By sweeping the gate voltage from positive to negative, the I-V curve is changed from highly nonlinear insulating behavior with a large gap to linear metallic behavior, and the linear-response conductance is increased by many orders of magnitude (see inset). The I-V characteristics indicate that the nanotubes are hole-doped semiconductors and that the devices behave as p-type FETs. 12

25 a b Figure 1.7 (a) Schematic figure of the experimental setup. A semiconducting nanotube is connected by two electrodes. Si substrate, which is covered by a layer of SiO 2, acts like a back gate. (b) Transport characteristics for a field-effect transistor employing semiconducting nanotubes. From Ref. [17]. E S 33 E S E M 11 E S Figure 1.8 Calculation of the energy separations E ii (d t ) for all (n, m) values as a function of nanotube diameter between 0.7 < d t < 3.0 nm (based on the work of Kataura et al. Ref.[18]). The crosses and open circles denote the peaks of semiconducting and metallic nanotubes, respectively. Filled squares denote the E ii (d t ) values for zigzag nanotubes, which determine the width of each E ii (d t ) curve. Note the points for zero gap metallic nanotubes along the abscissa. 13

26 1.4.2 Optical Properties of CNT The Kataura Plot Because of the very large electron density of states at the van Hove singularities (or subband edges), the intensity of the interband optical transitions E ii (d t ) is exceptionally strong, giving rise to exceptionally high intensities for the resonant Raman effect associated with the 1D density of electronic states. Thus the resonance Raman effect for carbon nanotubes can be expected to be much stronger than for 3D crystalline materials, based on the low dimensionality of carbon nanotubes. Figure shows the energies for the transitions between the van Hove singularities in the valence and condition bands E ii (d t ) of all possible (n, m) nanotubes. Figure is extensively used to interpret resonance Raman spectra in carbon nanotubes. Phonon Dispersion Relations for Nanotubes As a first approximation, the phonon dispersion relations for an isolated single-walled carbon nanotube can be determined by zone folding the phonon dispersion curves ω m 2D(k) of a two-dimensional graphene, where m = 1,,6 denotes the 3 acoustic and 3 optic modes and k is a vector in the layer plane. Since there are 2N carbon atoms in the unit cell, we will have N pairs of bonding π and anti-bonding π* electronic energy bands. Similarly, the phonon dispersion relations will consist of 6N branches resulting from a vector displacement of each carbon atom in the nanotube unit cell [13]. Figure 1.4.7(a) depicts the calculated phonon dispersion relations of an armchair carbon nanotube with C h = (10, 10). The number of degrees of freedom is 120, and the number of degrees of distinct phonon branches is 66. Figure 1.4 7(b) depicts the corresponding phonon density of states for a (10, 10) nanotube. Figure 1.4.7(c) shows a comparison between the phonon density of states g 1D (ω) for a (10, 14

27 10) nanotube shown as the solid curve and g 2D (ω) for a graphene sheet shown by the points [13]. Figure 1.9 (a) Calculated phonon dispersion relations of an armchair carbon nanotube with C h = (10, 10). (b) Corresponding phonon density of states for a (10, 10) nanotube. (c) Comparison between the phonon density of states g 1D (ω) for a (10, 10) nanotube shown as the solid curve and g 2D (ω) for a graphene sheet shown by the points. 15

28 Raman Modes of Carbon Nanotubes The special symmetry properties of 1D carbon nanotubes result in few Raman-active and infrared-active vibrational modes. Among the 6N calculated phonon dispersion relations for carbon nanotubes whose unit cell contains 2N carbon atoms, only a few modes are Ramanor infrared- (IR) active, as specified by the symmetry of the phonon modes. For example, Figure shows only the particular Raman-active modes that are expected to have significant intensity for the armchair nanotube (10, 10). Several of the mode frequencies and their Raman cross-sections are found to be highly sensitive to nanotube diameter, while others are not. Here it is seen that the radial breathing mode (RBM) A 1g, which occurs at about 165 cm -1 for an isolated (10, 10) nanotube (the figure (c) in the Figure 1.4.8), is strongly dependent on nanotube diameter, while the modes near 1580 cm -1 are not. Since this frequency is in the silent region for graphite and other carbon materials, this A 1g mode provides a good marker for specifying the carbon nanotube geometry. The frequency is inversely proportional to d t, within only a small deviation due to nanotube-nanotube interaction in a nanotube bundle. The equation to show the relationship between the frequency of RBM and the diameter of an isolated SWNT is ω 0 RBM (d t ) = ω 0 (10, 10) [d (10, 10) / d t ] ± (1.4.2) where ω 0 (10, 10) = 169 cm -1 and d (10, 10) = nm. This relation can be used as a secondary characterization tool for the diameter distribution in isolated SWNT samples. The Ramanactive E 2g mode (G band) of graphite at 1582 cm -1 corresponds to C-C bond stretching motions for one of the three neighboring bonds in the unit cell. In the Raman spectra of carbon nanotube samples, another peak (D band) around 1350 cm -1 is commonly seen. This peak has a character of disordered sp 2 bonded carbons, such as sp or sp 3 hybridized carbons. 16

29 The ratio of the intensity of E 2g peak (G band) over the peak around 1350 cm -1 (D band) is an indication of the purity of the sample. Figure is an example of the Raman spectra of a SWNT sample. 17

30 Figure 1.10 The calculated Raman mode atomic displacements, frequencies and symmetries for selected normal modes for the (10, 10) nanotube modes. The symmetry and the frequencies for these modes are not strongly dependent on the chirality of the nanotube. In the figure, we show the displacements for only one of the two modes in the doubly degenerate E 1g and E 2g modes.[19] 18

31 G RBM D Figure 1.11 Raman spectra of a SWNT sample. The inset figure shows the details of the RBM. The figure is from [20]. 19

32 1.4.3 Mechanical Properties of Carbon Nanotube Young s Modulus of Carbon Nanotube The carbon atoms of a graphene form a planar honeycomb lattice in which each atom is connected via a strong carbon-carbon sp 2 bond to three neighbouring atoms. Because of these strong bonds, the basal plane elastic modulus of graphite is one of the largest of any known material. Young s modulus (Y) for bulk graphite can be expressed as Y = C / h = 1.02 TPa, where h = nm and C = 342 N/m. For this reason, CNTs are expected to be the ultimate high-strength fibers. Besides the method involving Transition Electron Microscopy (TEM) [21], Atomic Force Microscopy (AFM) is also used to probe the mechanical properties of nanotubes. To minimize the uncertainty of the applied force, the spring constant of each AFM tip is calibrated by measuring its resonant frequency. The tip of an AFM is used to bend anchored CNT while simultaneously recording the force exerted by the tube as a function of the displacement from its equilibrium position. The value of Young s modulus value of singlewalled nanotube has been widely disputed due to different experimental measurement techniques. Some people have shown theoretically that Young s modulus depends on the size and chirality of single-walled nanotubes [22]. However, when working with different multiwalled nanotubes, others have noted that the modulus measurements of multi-walled nanotubes using AFM techniques do not strongly depend on the diameter [23]. Instead, they argue that the modulus of the multi-walled nanotubes correlates with the amount of disorder in the nanotube walls. Not surprisingly, when multi-walled nanotubes break, the outermost layers break first. Table lists the Young s modulus of several common materials for the purpose of comparison with carbon nanotubes. 20

33 Table Approximate Young s Modulus of Various Solids Material Young s Modulus in GPa Glass (all types) 72 Titanium (Ti) Carbon fiber reinforced plastic 150 (unidirectional, along grain) Steel Tungsten carbide (WC) Single carbon nanotube 1,000+ Diamond 1, Strength of Carbon Nanotube The theoretical calculation of a defect-free carbon nanotube provides limited prediction of real strength of CNT because the strength of a carbon nanotube is highly dependent on its structure. The nature of the defect is closely related to the synthesis process and absolutely defect-free carbon nanotubes are not easy to achieve by the current growth methods. Measuring the tensile strength of CNTs is also an extremely challenging task. Yu et al. [24] have conducted tensile testing of MWNTs. It was found that only the outermost layer breaks during the loading process. The tensile strength corresponding to this layer of CNT ranges from 11 to 63 GPa. The experiment setup is shown in Figure (a) and (b). 21

34 Figure 1.12 Tensile loading of individual MWNTs. (a) An SEM image of an MWNT attached between two AFM tips; (b) higher magnification image of the indicated region in (a), showing the MWNT between the AFM tips. The images are from [24]. 22

35 1.5 Applications of CNT Carbon nanotubes own many excellent properties, such as small size, tubular shape, high Young s modulus, high tensile strength, controllable metallicity, high electric current density transferring, high heat dispersion rate, roughness, and chemical inertness. Many researchers around the world try their best to realize their potential applications in many fields such as composites and fibers; sensors and probes; field emission devices; hydrogen storage media; and nm-sized semiconductor devices, and interconnects. But most proposed applications are still in the early development stage. Some prototypes are still in the laboratory instead of factory. This fact means there are still many challenges. Various challenges are associated with different applications. The main ones are: the properties of mass of CNTs are not as good as measured individual tube; high price of CNTs due to the low yield; properties depreciation on the interfaces between CNTs and traditional materials; lack of means to handle big amount of CNTs precisely; lack of means to sort mixed CNTs; production of CNTs with defects free and desired other properties; and so on. 23

36 1.6 Motivation and Overview We have seen that not only fundamental scientists favor carbon nanotubes for their perfect 1D molecular mode, but also applied scientists are enthusiastic about wide application of CNTs because of the potential shown from both theoretical calculations and current preliminary experimental results. At the same time, many factors associated with materials and techniques are hindering the rapid and broad application of CNTs. This work will focus on the growth mechanism and controllable synthesis of CNTs. The goals are to (1) achieve the ability to control the geometry of growing CNTs, (2) realize the low temperature growth of CNTs on glass substrate, and (3) evaluate the field emission properties of CNTs. In Chapter 2, we will describe how the Microwave Plasma Enhanced Chemical Vapor Deposition (MPECVD) system works to synthesize the vertical aligned CNTs. In Chapter 3, we will introduce a catalyst that was used to control the diameter of the nanotubes and discuss how to optimize the growth conditions. In Chapter 4, we will show a pretreatment to realize the low temperature growth of CNTs and explain the growth mechanism behind it. The field emission properties of the CNTs grown on glass substrate will be discussed in Chapter 5. Chapter 6 contains the summary. 24

37 Chapter 2. Synthesis of Carbon Nanotube 2.1 Introduction to Methods of CNT Synthesis This chapter introduces the four main methods of making carbon nanotubes electric arc discharge, pulsed laser ablation, thermal chemical vapor deposition (CVD) and plasma enhanced chemical vapor deposition (PECVD). The experimental setup, growth mechanism and virtues of each method will be included. The thermal CVD and PECVD will be emphasized Electric Arc Discharge In this method, two carbon rods are placed end to end, separated by approximately 1 mm, in an enclosure that is usually filled with inert gas (helium, argon) at low pressure (between 50 and 700 mbar). A direct current of 50 to 100 A driven by several tens of V creates a high temperature discharge between the two electrodes. The discharge vaporizes the anode carbon rods into carbon clusters, which are cooled to low temperature and form a small deposit on the other rod. Figure is the schematic drawing of the electric arc discharge experimental setup. The deposit contains carbon nanotubes, other nanoparticles and clusters. By optimizing the growth conditions, Ebbesen and Ajayan achieved growth and purification of high quality MWNTs at the gram level in 1992[1]

38 For the growth of single-walled tubes, the anode has to be doped with a metal catalyst such as Fe, Co, Ni, Y or Mo. A breakthrough in growing large amounts of SWNTs by arc discharge was made by Bethune and colleagues in 1993[25]. They used cobalt as the catalyst, which was added into the carbon anode and achieved a large amount of SWNTs in the soot material. Anode Plasma Discharge Cathode Figure 2.1 Schematic drawing of the electric arc discharge experimental setup. 26

39 2.1.2 Laser Ablation In 1995, Smalley's group at Rice University reported the first synthesis of SWNTs by laser ablation [4]. A schematic of the experimental setup is shown in Figure A pulsed laser is used to vaporize a graphite target in an oven at 1150 C. The ablation target was primarily composed of graphitic carbon, with a small percentage of cobalt and nickel catalyst. The oven is filled with argon gas in order to keep the pressure at 800 Torr. The laser beam is focused and scanned across the target surface during the ablation. A very hot vapor plume forms, then expands and cools rapidly at the end of the tube where chilling water pipes surround the tube. Soot material is collected at that part, and bundles of carbon nanotubes are obtained after the process of purification. There are many similarities between the electric arc discharge and laser ablation methods. The mixtures of carbon and catalyst in both methods are similar. Both methods involve the condensation of carbon atoms generated from evaporation of solid carbon sources at very high temperature, o C. The SWNTs and the MWNTs from those two methods are of excellent quality due to the high reaction temperature. The high quality and defect-free samples are very useful for the studying fundamental physics in low dimensional materials. But there are also some shortcomings of those two methods: 1. It is hard to scale up the yield of the carbon nanotubes. 2. Since the reaction is intense and occurring at very high temperature, it is hard to control. 3. To further integrate the carbon nanotubes into other devices, such as a field emission device or an atomic force microscope (AFM) tip, long and complex purification, filtration and cutting processes are needed. 27

40 Furnace Cooling end Laser Beam Plume Target Figure 2.2 Schematic drawing of the laser ablation experimental setup. 28

41 2.1.3 Thermal Chemical Vapor Deposition (CVD) Thermal chemical vapor deposition (CVD) is also known as catalytic chemical vapor deposition. The CVD method has been used to make carbon fibers since the 1970s. A schematic experimental setup is depicted in Figure In this method, Fe, Ni, Co or an alloy of the three catalytic metals is initially deposited on a substrate [26]. Sometimes there will be an underlayer of Mo or Al to optimize the growth of CNTs. Before the actual growth, there will be a pretreatment stage for the substrate, either by HF etching or by high temperature annealing in NH 3, H 2 or air to induce catalyst nm-size particle nucleation. The substrate then is put in the center of the furnace tube, where carbon source gases such as CH 4, C 2 H 2 or CO mixed with H 2, NH 3 or Argon are then introduced. To proceed with the growth reaction, the furnace temperature is increased to levels such as 500 o C-900 o C. The growth mechanism is similar to the growth mechanism developed in the 1970s [27, 28]. At high temperature, the carbonous gas dissociates into carbon atoms at surfaces of catalyst nanoparticles. More and more carbon atoms dissolve into catalyst nanoparticles and develop into metal-carbon supersaturated nanoparticles. The precipitation of carbon from the saturated nanoparticles leads to the formation of tubular carbon solids in a sp 2 structure. Depending on the position of catalyst particles after the growth of carbon nanotubes, there may be base growth when the catalyst particle resides at the base of the nanotube, or tip growth when the catalyst particle is at the tip. Figure shows the schematics of the proposed base growth and tip growth models. Sometimes a catalyst particle wrapped in the middle of a nanotube is also observed; since it constitutes neither base growth nor tip growth, we will denote it as middle growth. Figure shows one example. 29

42 The advantages of thermal CVD include: 1. Easy to scale up to industrial production. 2. Length and diameter of carbon nanotubes are controllable. 3. The patterned growth can be realized. Easy to integrate CNTs into electronic devices. 4. Simple experimental setups. But these advantages are mitigated by crystalline defects associated with the CNTs due to the low growth temperature of thermal CVD compared to laser ablation and arc discharge methods. Furnace Feeding gases Sample Figure 2.3 Schematic drawing of the thermal CVD experimental setup. Catalyst Carbon nanotube Tip growth Base growth Middle growth Figure 2.4 Schematic drawing of the three growth models. 30

43 Catalyst Nanotube Figure 2.5 HRTEM picture of one example of middle growth. The catalyst is neither at the tip nor at the base of the nanotube, but is wrapped in the middle of the tube. 31

44 2.1.4 Plasma Enhanced Chemical Vapor Deposition (PECVD) Plasma enhanced chemical vapor deposition (PECVD) can be regarded as a variation on thermal CVD. PECVD first emerged in the microelectronics industry because certain processes can t tolerate the high temperatures of thermal CVD. Relative to thermal CVD experimental setup, there is an additional plasma generation device in the PECVD system. The different means of plasma generation translate to direct current (DC) PECVD, radio frequency (RF) PECVD and microwave plasma enhanced chemical vapor deposition (MPECVD). MPECVD has been widely used in diamond growth for a long time and recently been adopted to fabricate CNTs [29, 30]. In a MPECVD system, microwave power typically is introduced by a waveguide through an input window, and plasma is generated by collisional electron-neutral impact as electrons are accelerated by the microwave electric field. During the discharge process, the carbonous gases dissociate into carbon atoms and radicals even before reaching the surface of the catalyst nanoparticles. Under the effect of a sheath electric field and diffusion, carbon species contact metal catalyst and go through the dissolvesaturate-precipitate process as mentioned above to form carbon nanotubes. Since part of the energy of carbonous gas dissociation is provided by the microwave power, instead of from thermal annealing exclusively as in thermal CVD, the minimum reaction temperature of MPECVD is not as high as in thermal CVD. Therefore, MPECVD has potential to be adopted for more types of substrates, such as glass and plastic. The total growth time is also decreased due to the presence of plasma. Another advantage of PECVD is the realization of vertically aligned growth of CNTs. Although vertically aligned CNTs have been achieved using thermal CVD [7, 9], the 32

45 mechanism of tube alignment of PECVD is different from that of thermal CVD. In thermal CVD, because of the high density, the Van der Waals attraction between nanotubes leads the tubes to grow closely together like towers, which supports the vertical alignment. In PECVD, the large electric field in the plasma sheath induces the alignment by providing dipole moments preferentially along the axes of carbon nanotubes, that act to align the tube in the direction of the electric field, which is perpendicular to the substrate in most cases [31, 32]. Therefore, the MPECVD is able to realize the vertically aligned and low density growth of CNTs on the substrate, which is very useful in the application of field emission devices involving CNTs[33]. 2.2 Experimental Setup for the MPECVD System The system was custom built. It contains microwave antenna and microwave guide; reaction chamber; temperature controlling units; gas flow controlling units and pressure controlling units. Figure is the schematic drawing of the system. Figure is an image of the actual system. All the units are manually controlled. Operators need to coordinate all the units to get the desired reaction parameters. 33

46 Reaction gases Microwave guide Microwave antenna Chamber Transformer Temperature controller Multi-channel gas flow controller Pump Pressure controller Figure 2.6 Schematic drawing of the MPECVD system built to fabricate carbon nanotubes. Figure 2.7 Image of the MPECVD system. 34

47 2.3 Parameters Affecting the Growth of CNTs by MPECVD System As research on carbon nanotubes has evolved for more than a decade since the most impacting discovery of CNTs by Iijima in 1991[1], researchers and engineers now have higher expectations for CNTs. Due to the efforts in understanding the growth mechanism and attempting to control the characteristics of as grown CNTs, more of these expectations are realized. In this part of my thesis, the experimental parameters to control the growth of CNTs, such as the type of the catalyst materials, feeding gases, effects of plasma and so on will be discussed Selection of Catalyst Material There is a wide selection of catalyst materials for the CVD that serves various purposes. But whatever the form of catalyst employed before the growth, the real catalyst during the CNT growth involves transition metals (Fe, Co or Ni) and their alloys[26] in the form of nanoparticles, due to the growth mechanism mentioned above. Here are some published examples of selecting a specific catalyst to achieve specific objectives. To realize continuous growth of CNTs that can increase the industrial production yield, researchers at Rice University reported a continuous-flow synthetic, high pressure thermal CVD utilizing Fe(CO) 5 as the catalyst for the gas phase [34]. Another example of continuous addition of catalyst material is reported by G. Eres and colleagues in 2004 [35]. Setting aside the commonly seen tri-layer of Al (10nm), Fe (1nm) and Mo (0.2nm), another catalyst Fe(C 2 H 5 ) 2 can be dissolved in hydrocarbon solvent and sent to the reaction chamber continuously. Its vertically aligned CNT film can be as long as 3.25 mm. 35

48 During this study, I chose two catalysts: iron-containing block copolymers and pure Fe film. In later chapters, I will focus on the experiments and results from those two catalysts Selection of Feeding Gases The feeding gases in the CVD method generally consist of carbonous gases such as hydrocarbon gas (C x H y ) or carbon monoxide (CO) and balance gases such as nitrogen (N 2 ), hydrogen (H 2 ) or ammonia (NH 3 ). In MPECVD the ratio of carbonous gas to the total gas is in the range of 1/10 to 1/4. If the ratio is too high, CNTs can t be formed, since carbon atoms won t dissolve in the catalyst metals due to the barrier, an effect called catalyst poisoning. If the ratio is too low, effective growth won t be achieved. So the ratio of feeding gases always needs to be optimized for the particular CNT growth experiment. Including a specific carbonous gas, such as CO, is another method to minimize the formation of amorphous carbon [34]. A recent discovery is that amorphous carbon can be oxidized by adding oxidizer during thermal CVD growth. Researchers from Japan[36] and Stanford University [37] added small amounts of H 2 O and O 2 during the growth and observed very efficient growth of SWNTs due to the oxidizing effect of those two gases. In MPECVD, NH 3 and its radicals are believed to have efficient etching effect on amorphous carbon, so these are commonly used in MPECVD as balancing gases Effect of Plasma and Sheath Electric Field on CNT Growth in MPECVD In MPECVD, the plasma and the sheath play important roles in the formation of CNTs. When the feeding gas ratio has been set, the gas pressure, forward microwave power and 36

49 reflection microwave power have to be adjusted carefully to ignite and maintain plasma. The gas pressure is monitored by a cold cathode gauge, and the exhausting valve is adjusted according to the difference between the actual value and the set value. The forward microwave power is controlled by a knob on the control panel of the microwave producer. The reflection microwave power is adjusted by moving the shorting plunger and adjusting the three stub tuners. Figure (a) shows details of the core structure of the reaction chamber. Figure (b) shows the boundary between the plasma and the substrate. In plasma accelerated electrons, impact leads processes of excitation and ionization. There are five reactions known to occur [38]: 37

50 Description: 1. The three stub tuner adjusts the path of the microwave. 2. The direction of the traveling microwave inside the microwave guide. 3. The gas inlet route. 4. The plasma ignited by the microwave. 5. The substrate with catalyst film on. 6. The shorting plunger to control the reflection of the microwave. This and the turning knobs control the forwarding and reflection power of the microwave. 7. The gas exhausting line. Figure 2.8 (a) Schematic drawing of the details of the system s reaction chamber. 7 Plasma Excitation, Ionization, Dissociation, Radiation, Recombination Neutral species Ionic species Boundary Sheath Surface Substrate Diffusion Acceleration Migration, Ion bombardment Reaction adsorption dissociation Desorption Chemicalsputtering Figure 2.8 (b) Boundary between the plasma and the substrate [39]. 38

51 1. Excitation A + e A * + e AB + e AB * + e 2. Dissociation AB + e A + B +e 3. Direct ionization A + e A + + 2e AB + e AB + + 2e 4. Cumulative ionization A * + e A + + 2e AB * + e A + + B +2e 5. Dissociative ionization AB + e A + + B +2e Here e denotes the electron; A, B, or AB the atom or molecule; and A * or AB * the excited atom or molecule. (The reactions lacking electron involvement are not listed here.) According to the computational model developed by Meyyappan and colleagues[33], the plasma of C 2 H 2 and NH 3 contains 21 kinds of neutral species, 7 kinds of charged species and 200 reactions. Unlike the thermal CVD where the energy of dissociation of hydrocarbon gas originates only from external heating, for MPECVD part of the dissociation energy comes from the microwave energy. Therefore the reaction temperature can be lower than in the thermal CVD. The annealing temperature is lower also because the ion bombardment is intense enough to elevate the temperature of catalyst islands. Since inside the plasma, electron density is equal to ion density, the plasma is equipotential, which implies no electric field inside the plasma. Between the plasma and the interfaces that include the surface of the substrate, there is a boundary area called sheath. 39

52 Besides the moving particles, there is an electric field existing in this area. Because the current density is determined by the speed of the electrons and ions, and the average speed of electrons is far greater than that of ions, the surface of the substrate accumulates more electrons than ions[38]. The potential of the substrate (V s ) is lower than the potential of the plasma (V p ). Figure shows the sheath electric field mentioned above. The difference between V p and V s is about 10 to 15 V. The length of the sheath is about 100 µm[38]. It is believed that in MPECVD, the sheath electric field is perpendicular to the surface of the substrate[40]. The author of this article also proposed several CNT growth geometries, which are shown in Figure Vp Potential x Plasma Sheath Vs Substrate Figure 2.9 Schematic drawing of the sheath electric field. 40

53 Figure 2.10 (a) SEM micrograph showing the radially grown nanotubes on the surface of a 125-µm-diameter optical fiber. (b) Close-up micrograph showing the conformally perpendicular nature of the nanotube growth on the fiber. (c) (f) Examples of nonplanar, complex surfaces where nanotubes can be conformally grown perpendicular to the local surface. This figure is from[40]. Figure (f) shows that the growth direction of CNTs at both the tops and roots of the pillars are coincidently outward and perpendicular to the surface of the pillars. But the author didn t consider the change of the direction of the electric field due to the connection of the vertical wall and horizontal floor. At lower part of the pillar, the electric field bends upward when it is away from the surface of the pillar. Since the growth direction of CNTs would follow the direction of the electric field, we found up-bending CNTs on this kind of substrate. Figure is the SEM picture of the carbon nanotubes from the bottom part of 41

54 a vertical surface to show the shape of the CNTs. Figure is the simulation of the distribution of the ambient electric field and potential around the step-like protrusion on the substrate. By comparing these two figures, we can conclude that the direction of the CNT growth closely follows the direction of the ambient electric field on the substrate. Therefore some non-straight as-grown nanotubes can be achieved, and the discovery of some interesting shapes of as-grown CNTs confirms our understanding of the mechanism of the alignment of CNTs by MPECVD. The electric static force induced by the electric field is not the only factor affecting the geometry of the CNTs. The following example illustrates another two factors that determine the shape of as-grown CNTs. Figure shows a square pattern of CNTs. The surrounding CNTs are longer and denser than at the center. This is due to the outward migration of the Fe film at the edge of the pattern during annealing. This migration creates a ramp of Fe film on the edges of the pattern. The ramp part is thinner than the central part of the film, which makes it easier and quicker to form nanoparticles during the annealing and plasma etching. The schematic figure of the formation of the Fe ramps is shown in Figure Therefore, the growth of CNTs on the surrounding part will be one step ahead of the central part, which makes the CNTs longer. The electric field above the surface should be perpendicular. So the lower parts of CNTs are vertically aligned to the surface. But this doesn t explain why the tops of all the CNTs bend toward the center of the square. This bending occurs because of the large difference in growth rate of central CNTs versus surrounding CNTs, as well as the Van der Waals attraction force, which keeps CNTs close to each other. Since the central CNTs are shorter than the surrounding CNTs, the only way for CNTs to bear the force is to bend. Therefore, at that height of the CNTs, the electrostatic 42

55 force from the electric field is no longer dominant. In summary, by figuring out the factors that affect the growth of CNTs, we may be able to engineer the growth of CNTs by designing the geometry of the substrate, thickness of the catalyst and length of the CNTs. 43

56 Figure 2.11 SEM picture of showing the shape of the CNTs extending out from the vertical surface at the bottom of the object. Figure 2.12 Distribution of the electric field and potential around the step-like protrusion on the substrate. 44

57 Figure 2.13 SEM image showing a square pattern with CNTs. The surrounding CNTs are longer and denser than those in the center. The central part of the Fe pattern The ramp of the Fe film Figure 2.14 Schematic figure of the formation of the Fe ramp film on the edges of each pattern. 45

58 2.4 Conclusion In this chapter, the four CNT growth methods of laser ablation, arc discharge, thermal CVD and MPECVD were introduced. Later we focused on the MPECVD. The description of the home-assembled MPECVD system was given. The reason that the MPECVD s capability of fabricating CNTs at low temperature was explained. The reactions occurring in the boundary between the plasma and the substrate and the sheath electric field were discussed. The up-bending as-grown CNTs were observed and the explanations were given. It was found that factors including the induced sheath electric field, catalyst film thicknesses as well as Van der Waals force play important roles in determining the geometry of as-grown CNTs. The study showed the potential of producing CNTs with desired shape by manipulating the above factors. 46

59 Chapter 3. Fabrication of Small-diameter CNTs Using the Polymer Based Catalyst In this chapter I will report on a method to produce CNTs by using metal-containing diblock copolymers. The polymer serves as a template to control the number of iron atoms per nanoparticle and the spacing between the nanoparticles. Therefore, the diameter of CNT and the spacing are controllable. By carefully optimizing the pretreatment and MPECVD growth conditions, the growth temperature can be as low as 604 o C. The outer diameter of CNT is consistently in the range of 3 to 6 nm. There are two to six wall layers. The methods to control of the density and the length of the as-grown nanotubes are discussed. 3.1 Introduction Among the various methods to grow CNTs, plasma enhanced chemical vapor deposition (PECVD) shows particular promise. PECVD has shown the advantages in producing vertically aligned CNTs [41] and growth of CNTs at low temperature [42]. The widely used catalyst for PECVD method is transition metal (Fe, Co and Ni) film deposited by either thermal evaporation or electron beam deposition. But the continuous metal film won t help in catalyzing the growth of CNTs. Before the flowing the hydrocarbon gas, the metal film breaks into small catalyst islands due to the high temperature annealing and/or plasma ion bombardment. The size of the catalyst islands determines the size of the as-grown CNTs. Researchers have reported that the size of the catalyst island and consequently the size of the

60 CNTs, are mainly determined by the thickness of the original metal film [40]. At the same time, the sizes and the size distribution of the catalyst islands are also determined by the pregrowth treatments. One group of researchers used hydrofluoric acid dipping to increase the surface roughness of the catalyst film. By dipping the silicon substrate with Ni film as the catalyst into the hydrofluoric acid, the Ni catalyst film is etched thinner and cracks are produced. They also found the relationship between dipping time and the growth of CNTs [43]. The addition of metal underlayer such as Al or Ir also helps to increase the density of the growth of CNTs [44]. The author attributed the increase in the density of the CNTs to the increase in the surface roughness and more active nucleation site density provided by the proper thickness and the proper metal of the underlayer film. In a PECVD system, the plasma can be used to etch the surface of the catalyst film to make it rough. The intensity of the plasma etching is determined by many factors. But the results of breaking the continuous film into nanoparticles seem similar. NH 3, N 2, and H 2 have been used to produce plasma to etch catalyst film [41, 45]. Under the electric field existing in the plasma sheath, the ionized species are accelerated to strike on the catalyst surface. The energy delivered to the catalyst film by the striking increases the kinetic energy of the catalyst atoms, which increases the temperature of the film. The film finally breaks into islands due to the surface tension. But due to the imperfect metal film deposition at a µm or even nm scale, uneven tension during the film break and interaction between the catalyst film and the supporting surface, the size of the broken catalyst islands can t be consistent to nm scale. The non-uniformity of the catalyst islands leads to growth of non-uniform diameter CNTs. There are also some attempts to minimize the diameter of as-grown carbon nanotubes and reduce the size differences in fabricating SWNT by thermal CVD method. One example 49

61 is to use 0.1 to 0.2 nm Fe film as the catalyst film [46]. Considering the diameter of an iron atom is nm, the thickness of the film used in here is almost the thinnest film people can achieve. Another example is the use of metal salt solution as the catalyst[20]. But none of these methods are able to control the density of nanotubes or control the diameter and number of walls of the as-grown carbon nanotubes. A new technique to fabricate uniform sized, controllable density and diameter CNTs is needed. The structure of the metal containing copolymer catalyst determines that it is a promising catalyst for the growth of CNTs. Each chain of the copolymer can be divided into metal containing component and non-metal containing component. Every chain connects other surrounding chains. Thus, the matrix of the metal-containing copolymer serves as a template for the catalyst. The size of the metal-containing micelle determines the number of iron atoms inside, and the length of the non-metal containing chain determines the spacing between each metal-containing micelle. By carefully adjusting the copolymer composition, catalyst size and spacing can be controlled on the nm scale. By using this kind of catalyst, controllable diameter with narrow diameter distribution and controllable spacing vertically aligned CNT array can be grown via MPECVD. 3.2 Catalyst Preparations Our work used diblock copolymers of amorphous polystyrene-blockpolyferrocenylethylmethylsilane (PS-b-PFEMS). Figure shows the formula of the catalyst. With a PFEMS volume ratio of 33%, the value of m and n in the formula are 402 and 98, respectively. The copolymer chains connect to one another to form the spheres of 50

62 PFEMS in a matrix of uniform PS. The catalyst solution is spin-coated onto the surface of the silicon wafer to produce a well-dispersed monolayer. The domain size of PFEMS is about 22 nm with a spacing of 36 nm. Figure is the AFM image of the catalyst layer. 51

63 Et Figure 3.1 Molecular formula of the iron-containing diblock copolymer unit. Figure 3.2 AFM image of the surface of the substrate after the spin-coating of the catalyst. Dark regions are PFEMS while lighter regions are PS. 52

64 3.3 Experimental Setup Depending on the means of igniting and maintaining the plasma, PECVD can be subdivided into DC (Direct Current) PECVD, RF (Radio Frequency) PECVD, MPECVD, and so on. A microwave antenna is used to produce microwave at the frequency 2.45 GHz. The maximum power of the microwave is 1.2 kw. Besides the microwave power, there is a separate heating system that can heat the substrate to 900 o C independently. There is also a rotary pump that can drive the system down to 1x10-3 Torr. The diameter of the reaction chamber is 12.5 cm. The sample stage is a molybdenum disk with a hole in the center. The thermal couple tip is set inside the hole and attached to the bottom of the sample. There are three mass flow controllers that control the flow speed of three feeding gases independently. The SEM pictures were taken by a Hitachi S-4700 Field Emission Scanning Electron Microscope. The High Resolution Transmission Electron Microscopy (HRTEM) pictures were taken by the JEOL 2010F-FasTEM. The ion-reactive etching system is a SAMCO Model RIE-1C Reactive Ion Etcher. The Raman spectra were recorded using a Dior XY triple spectrometer and collected using a charge-coupled device (CCD) cooled with LN Catalyst Pretreatment In this step, the substrate with a layer of catalyst was etched by O 2 plasma in the reactive ion etching chamber. The flow speed of O 2 was 50 standard cubic centimeters per minute (sccm). The pressure inside the chamber was 245 mt, and the temperature was room temperature. After a 1 m treatment of the O 2 plasma, the organic element was removed and the iron was oxidized into Fe 2 O 3 nanoparticles. We had previously tried using ultraviolet light ozone treatment to remove the organic elements, but this elevated the reaction 53

65 temperature, causing the nanoparticles to aggregate into big clusters that would only produce large diameter CNTs. Therefore, we used the O 2 plasma at room temperature. Fig 3.4.1(b) shows the SEM image of the catalyst surface after the O 2 plasma treatment. The image illustrates the Fe 2 O 3 in the form of separated nanoclusters without coalescing and with a very clean surface. We also explored the effects of O 2 plasma etching time on the catalyst. Fig 3.4.1(a) and (c) are the SEM pictures of the substrate surface after 0.5 m and 3 m etching in the O 2 plasma, respectively. From Fig 3.4.1(a), we see that the 0.5 m O 2 plasma etching is not sufficient to remove the organic micelle completely, so that there are too few reflection electrons to produce an SEM picture with good contrast and sharp edges. From Fig 3.4.1(c), we see that the diameter of each catalyst island produced in 3 m etching is 2 to 3 times larger than the catalyst etched for 1 m. We suspect that the long time of ion bombardment on the surface increases the surface temperature, which enables the aggregation of the nanoparticles. The optimized treatment time of O 2 plasma is 1 m. 54

66 a b c Figure 3.3 SEM pictures of the catalyst after the substrate has been treated by the O 2 plasma. (a) 0.5 m, (b) 1 m, (c) 3 m. The white dots are the Fe 2 O 3 molecules. 55

67 3.5 Growth procedures Generally, there were three steps in the CNT growth process. In step 1, the O 2 plasma etched substrate was transferred in air to the growth chamber. The chamber was pumped down to the base pressure of 1x10-3 Torr. Then a gas mixture of 10 sccm H 2 / 190 sccm N 2 was introduced. To see the effect of pressure on the growth of CNTs, the pressure was increased from base pressure to 23 Torr, 27 Torr and 30 Torr for different trials while keeping other parameters constant. At the same time, the heating system heated the substrate to the desired temperatures. To see the effect of temperature on the growth of CNTs, we tried temperatures of 885 o C, 825 o C, 660 o C and 600 o C while keeping other parameters constant. When the desired pressure and temperature were achieved, the system was left at this status for a couple of m in order to reduce the iron oxide to iron by H 2. In step 2, the microwave was transmitted to the chamber with a power of 500 W. The mixture of H 2 and N 2 gas was stricken into plasma. Consequently, a gas mixture of H 2 and N 2 was stopped and NH 3 flew in at speed of 150 sccm. The NH 3 plasma was on for 2 m. Most iron oxide nanoclusters were reduced to iron nanoparticles during this step. The molecules and radicals of NH 3 reduce the iron oxide very effectively. In step 3, the gas mixture of NH 3 and C 2 H 2 was added to the chamber. The growth time was 3 m. The iron oxide was further reduced and CNTs were deposited on the surface of the substrate. To see the effect of gas ratio on the growth of CNTs, different ratios were tested while keeping the other parameters constant. 56

68 Table 2 Growth Parameters for the Samples Reported Sample Pressure (Torr) Temperature ( o C) C 2 H 2 /NH 3 (sccm) No /150 No /150 No /150 No /150 No /150 No /150 No /150 No /150 No /150 No / Results Effect of Growth Temperature In this part of the experiment, the following parameters were kept constant: pressure was 26.8 Torr, flow rate of NH 3 was 150 sccm during the NH 3 etching part, and flow rate of C 2 H 2 was 50 sccm and NH 3 was 150 sccm during the growth part. The growth temperatures were 885 o C, 825 o C, 660 o C and 600 o C (sample numbers 1 to 4 in Table 1). Figures (a) (d) are the SEM pictures of the CNTs grown at those temperatures. The pictures show that the length of the nanotubes decreased from a few µm to 0.7 µm and so did the density when temperature decreased. The decreasing of the length and density of the nanotubes produces a 57

69 low growth rate of the CNTs, which may be due to the low decomposition rate of the C 2 H 2 [47] and less activity of the catalyst at low temperature. The SEM pictures also show that the diameter of the CNTs didn t change when temperature varied. Some studies have found the diameter of the CNTs to increase when the temperature decreases [42], while others have found the diameter to decrease [33, 48]. This is because the formation of nano-size catalyst islands is highly related to the growth temperature in those methods. In our method, the size of the catalyst islands was predefined and remained unaffected by variation in the growth temperature. 58

70 aa ab c d Figure 3.4 Effect of temperature on the growth of CNTs. The higher the temperature, the longer and denser the CNTs grown. (a) 885 o C; (b) 825 o C; (c) 660 o C (d) 600 o C. 59

71 3.6.2 Effect of Growth Pressure In this part, all the parameters were kept constant except for pressure. The growth temperature was 600 o C, ratio of NH 3 to C 2 H 2 was 150/50 (sample numbers 4 to 6 in Table 1). The SEM pictures of (a), (b) and (c) are for the pressure of 23 Torr, 27 Torr, and 30 Torr, respectively. We see that when the pressure is in the range of 20 Torr to 30 Torr, the density and the average length of the CNTs increases with the increase of the precursor gas pressure. We believe that higher pressure can provide larger amounts of feedstock gases. The decomposition of hydrocarbons increased, and more carbon were produced to supply more CNTs growing. We didn t observe more amorphous carbon being produced when the pressure was increased, contrary to what has been reported[48]. When the pressure was over 30 Torr, it was hard to ignite the plasma because the mean free path of electrons was too small for the plasma to sustain. Another reason for the higher growth rate of carbon nanotubes with increasing of the pressure is due to the decrease in plasma etching. The electric field in the plasma sheath zone won t change with the change of the pressure. The higher pressure reduces the mean free path of ions which leads to the smaller average kinetic energy of traveling ions. The bombardments to the carbon nanotube brought by the high speed ions will be sequentially reduced. Therefore, the etching speed of carbon nanotubes is reduced. So the higher pressure results in higher supply of carbon atoms and less etching of carbon nanotube by plasma. The density and length of as-grown carbon nanotubes are increased. 60

72 a b c Figure 3.5 SEM pictures showing the effect of pressure on the growth of CNTs. (a) 23 Torr, (b) 27 Torr, (c) 30 Torr. 61

73 3.6.3 Effect of Feedstock Gas Ratio Another significant factor in the growth of CNTs is the gas ratio. To test it, we did the following experiments. We kept every parameter constant except for the gas ratio. The temperature was 600 o C and the pressure was 30 Torr. Figs (a) (d) (sample numbers 6 to 9 in Table 1) are the SEM pictures showing the results. The ratios of C 2 H 2 over NH 3, from left to right, were 15/150, 30/150, 50/150, and 75/150 sccm. The effect of the feedstock gas ratio is evident. The high concentration of C 2 H 2 was found to hinder the growth of CNTs. Figure (d) shows hardly any CNTs on the surface. Fig is the Raman spectrum of those four samples. The figure shows two peaks centered at 1,350 (the D band) and 1,598 (the G band). As the ratio of C 2 H 2 over NH 3 was decreased, the shapes of both peaks became sharper and the intensities became stronger. The well defined-shape of the G band indicates the well-defined graphene morphology of the CNTs; the formation of the D band is due to the amorphous carbon. Therefore, we can conclude that the decrease in the ratio of C 2 H 2 over NH 3 favors the formation of the graphene layer on the surface of the saturated catalyst islands. The reason is because of the etching effect of NH 3 and NH 3 radicals. When the gas ratio of NH 3 is decreased, the amorphous carbon formed on the surface of the catalyst islands won t be etched away which will block the entrance of new carbon atoms to the catalyst islands. The growth of carbon nanotubes turns to worse. The lower flow ratio of C 2 H 2 /NH 3 can t be achieved due to the limitation of the device. The best ratio is 15/

74 a b c d Figure 3.6 SEM pictures showing the CNTs grown at different gas ratios of C 2 H 2 /NH 3. (a) 15/150 sccm, (b) 30/150 sccm, (c) 50/150 sccm, (d) 70/150 sccm. 63

75 Figure 3.7 Raman spectrum of the samples with different gas ratios. From top to bottom: 15/150 sccm; 30/150 sccm; 50/150 sccm and 70/150 sccm of C 2 H 2 /NH 3. 64

76 3.7 Characterization High Resolution Transmission Electron Microscopy (HRTEM) The SEM pictures show that the diameter of the as-grown CNTs is very consistent even when the growth conditions are varied. To determine the inner structure of the CNTs, HRTEM was performed on sample number 7. We used the TEM copper grid to scratch the surface of the substrate. The CNTs were caught by the grid and the carbon film on the grid. Figures (a) and (b) are high magnification HRTEM images, which reveal that most CNTs have two to six wall layers. Figure (c) is the diameter distribution of as-grown carbon nanotubes. The diameter is consistently in the range of 3 to 6 nm. There is no bamboo-like structure or stacked-cone structure, even though these have been commonly seen on CNTs grown by PECVD method [36, 42, 48, 49]. We found no catalyst when we took the TEM pictures. Fig is the SEM picture showing the area scratched. We can see the catalysts left on the surface. We also used an acetone ultrasonic bath to remove the as-grown CNTs. Fig is the SEM picture after the bath. There are no CNTs left on the surface, only catalysts. So, we can conclude that the CNTs we produced were base-growth CNTs as reported in the literature [40]. 65

77 6 c 5 Number of counts Outer Diameter range (nm) Figure 3.8 (a) and (b) are HRTEM images of sample number 7. (c) The diameter distribution of the as-grown carbon nanotubes. 66

78 Figure 3.9 SEM picture of the area scratched by the TEM grid. The CNTs are gone. Only the catalysts remain. Figure 3.10 SEM picture of the substrate after the acetone ultrasonic bath. The CNTs are washed away. Only the catalysts are left on the substrate. 67

79 3.7.2 Vertically Alignment It is well known that the electric bias imposed on the substrate is the reason to align the CNTs grown via MPECVD [40, 50]. Vertically grown CNTs have also been produced by thermal CVD [36, 51]. The CNTs are densely packed and held together by Van de Waals interactions [32]. The high density of the CNTs and the fast growth rate are the key motivations for achieving vertically aligned CNTs grown by thermal CVD. When a vacuum leak causes the large amount of oxygen or water to enter the reaction chamber, this has been shown to poison the catalyst. Therefore, the growth rate of the CNTs is decreased, and the CNTs are not vertically aligned [20]. The method of MPECVD can realize vertical alignment of CNTs at low growth density. Fig is the SEM image of the cross section of sample number 10. It shows that most of the CNTs stand up from the substrate. Fig is the SEM picture showing the CNTs on the surface after scratching by the TEM grid of sample number 2. In contrast to Figure (b), the CNTs are lying on the surface due to the scratch. The vertical alignment tendency of our small diameter CNTs is very good. 68

80 Figure 3.11 Side view SEM picture of sample number 9. The CNTs are vertically aligned to the substrate surface. Figure 3.12 SEM picture of the surface of sample number 2 after the scratch by the TEM grid. In contrast to Fig 5.1.1(b), the CNTs are lying down due to the scratch. 69

81 3.8 Conclusion By using the iron-containing diblock copolymer catalyst, we can realize the growth of small diameter CNTs with good inner structure and standing tendency at as low a temperature as 600 o C via MPECVD. The growth parameters such as temperature, pressure, and gas ratio were studied and the optimized growth conditions were found. The CNT diameter is consistently in the range of 3 to 6 nm and consistent with the diameter of the catalyst. There are two to six wall layers. The diameter distribution of the CNTs is narrow due to the intrinsic property of the catalyst. To further decrease the growth temperature and integrate the CNTs into device might be the next approach. 70

82 Chapter 4. Low Temperature Growth of CNTs Using Iron Film Catalys 4.1 Introduction Carbon nanotubes (CNTs) have been regarded as a promising material for applications in various fields due to their special structure, unique dimension, strength, flexibility, thermal properties and unique electronic and magnetic properties. For example, SWNTs have the highest current-carrying capacity of any conductor, which makes them an attractive electronic, sensing, or heat dispersing material for nano-electromechanical systems (NEMS) and future integrated circuitry [52]. Researchers have succeeded in making a CNT-based transistor where the CNT is made by laser ablation and is transplanted to the gap between the source and drain contacts [17]. For factory manufacturing in the future, it is ideal to have in situ growth in IC. The current complementary metal oxide semiconductor (CMOS) technology allows a maximum processing temperature of o C due to the mechanical integrity of low dielectric constant intermetal dielectrics [53]. Another example of a CNT application requiring low temperature is making cathodes out of CNTs for display. The substrate material is glass whose glass transition temperature is 550 o C. CNTs are considered candidate material for the cathodes due to their small size, tubular geometry and high stability. There are several methods to integrate CNTs into the cathode substrates screen printing, electrophoretic deposition (EPD) and CVD. Figure and Figure show the

83 schematic figures of EPD and screen printing, respectively. Both of these require cleaned and filtered CNTs made beforehand by synthesis methods and a series of purification steps. During EPD, the CNTs are mixed with ethanol and charger. The substrate with masks affixed is attached to the anode or cathode (determined by the polarity of the charger) and CNT film is deposited by direct current [54]. 72

84 E Description: 1. The counter electrode of the substrate cathode. In this case the anode. 2. Carbon nanotubes attached with carrier ions to be moving to the deposition area. 3. The carrier ions which move to the substrate under the effect of electric field. 4. Mask with open area which to be deposited by CNTs film. 5. Deposited CNTs film on the substrate. 6. The substrate, in this case acting the cathode. 7. The insulated substrate holder. 8. Liquid media, usually organic liquid. Figure 4.1 The schematic drawing of the EPD method A Description: A: The process of the printing. B: The substrate with CNT pattern film after the removal of the mask (3). 1. Printing wheel with paste (2). 2. The printing paste which is the mixture of CNTs and organic paste. 3. Mask laying on the substrate. 4. The substrate. 5. Patterned CNTs film. Figure 4.2 The schematic drawing of the screen printing method. B 73

85 One shortcoming of this method is that the very small depositing pattern is hard to achieve, because the long and tangling CNTs will attach to the edge of the small opening mask and the accumulating chunk of CNTs stops the channel through which CNTs would enter. Another is that the charger salt may reduce the lifetime of the cathode. A third shortcoming is that the whole substrate requires submersion in liquid ethanol, whose effects on other electronics in the substrate are unknown. In the screen-printing method, the pre-produced carbon nanotubes are purified and mixed with epoxy/binder, and then screen printed or applied at required locations. The first carbon nanotube display was realized by this method [55]. One shortcoming of this method is that the hole size has to be bigger than 10 µm in order to print the paste effectively [56], making the resolution difficult to increase. Another shortcoming is from the organic paste, which is believed to be responsible for arcing while the cathodes are in operation. To overcome the shortcomings of the methods mentioned above and to realize the promising applications of CNTs requires in situ growth of CNTs by CVD, which at the same time is able to pattern the growth easily. Because using high temperatures for CNT growth is destructive to many substrate materials, the technique of low temperature growth of CNTs would expand the range of CNT applications. A lot of efforts have been devoted in the past to lower the growth temperature for CNT. Many of the conditions required for low temperature growth have been discovered. Back in 1998, Ren et al. succeeded in making vertically aligned CNTs on glass by using plasmaenhanced hot filament CVD [30]. The growth temperature is as low as 666 o C. Researchers in Japan have succeeded in making CNTs by radio frequency (RF) PECVD on glass substrates [57]. They found that the magnitude of the additional DC bias voltage applied between top 74

86 and bottom electrodes determines the collision energy of carbonic cations with substrates, which determines the temperature of the substrates. By carefully adjusting the DC bias voltage, the growth of CNTs on glass substrates is realized. Another example of succeeding in growing carbon nanofibres on glass substrates by direct current (DC) PECVD is from researchers at Boston College. Their work focuses on the relationship between plasma current and carbon nanofibres growth [41]. But little research has been done to explore the relationship between growth temperature and catalyst treatment in PECVD. By calcining in air pretreatment and oxidizing under oxygen plasma of iron catalyst film, we also are able to realize the growth of CNTs on glass substrate. We now describe this method and explore the effects of catalyst treatment on low temperature growth of CNTs. 75

87 4.2 Experiment There are two sequential parts to the experiment. The first one concerns a silicon substrate and the second one concerns a glass substrate. Additionally, the results of the first part direct the processes of the second part. Treatments of Air Annealing (Sample A) and Oxygen Plasma (Sample B) For Sample A of silicon substrate, the catalyst was 5 nm thick iron film deposited by thermal evaporation. Then the substrates were annealed in air at 500 o C for 3 h for the oxidation of Fe to Fe 2 O 3. After the pretreatment, the substrates were transferred to the MPECVD chamber. The pressure was reduced to 1x10-3 Torr by the rotary pump. When the base pressure was achieved, a gas mixture of 10 sccm H 2 / 190 sccm N 2 was introduced and chamber pressure was increased to 30 Torr. At the same time, the temperature of the substrate was increased to 500 o C, which was kept constant for the whole experiment. After 10 m to stabilize the system, a 500 W microwave was produced and transmitted to the chamber and H 2 /N 2 plasma was ignited. When the H 2 /N 2 plasma was ignited, the NH 3 flowed into the chamber in the flow rate of 150 sccm and H 2 /N 2 was shut off. The system was left in this state for 3 m to allow the partial reduction of the Fe 2 O 3 to Fe by NH 3 and radicals. Then the pressure was increased to 35 Torr and microwave power was increased to 700 W, these being the appropriate values for growth. This state was maintained for 3 m to enable NH 3 plasma reduction. Then 50 sccm C 2 H 2 was introduced to the chamber. The gas was a mixture of NH 3 and C 2 H 2, and the color of the plasma ball turned to blue and purple. The growth time was 3 m, and the substrates were taken out for further characterization and testing. Figure 4.2.1(a) is the SEM image of the as-grown CNTs. 76

88 To grow small-diameter CNTs, the thickness of the Fe film of Sample B was decreased to 0.2 nm. The growth parameters were set to the optimal values. The details on optimizing the growth conditions can be found in Chapter 3. Instead of annealing the substrate in air, we opted to oxidize the catalyst using oxygen plasma for 5 m, which produces Fe 2 O 3 as effectively as calcination in air at 500 o C for 3 h. The biggest advantage of O 2 plasma oxidization is that the reaction occurs at room temperature and process time is shorter. In summary, here are the conditions of Sample B that are different from Sample A shown in Figure 4.2.1(a): the thickness of Fe is 0.2 nm, the treatment of the Fe film is O 2 plasma oxidization for 5 m, the growth pressure is 42 Torr, the flow speed of C 2 H 2 is 30 sccm, the microwave power is 580 W and the growth time is 1.5 m. Figure 4.2.1(b) shows the SEM picture of the as-grown CNTs after the above changes. Figure is the HRTEM image of the same CNT sample to show the diameter and the crystallization of the as-grown CNTs. From the images, we can see that (a) the oxygen plasma treatment has the same effect as air annealing treatment; (b) the pretreatment of oxidization is good for thinner catalyst film as well; (c) the density of the CNTs is greatly increased; (d) the diameter of the CNTs from thinner Fe film is 6-8 nm. Growth of CNT on Soda-lime Glass Substrate Based on the results from the part of the experiment that used silicon substrates, the sodalime glass substrates were also used to allow the growth of CNTs successfully. A 2.5 cm by 2.5 cm aluminum foil with a square hole of 1 cm by 1 cm was used as a mask for the 5 nm thick Fe thermal deposition to get the pattern of Fe catalyst film. The pretreatment is O 2 plasma oxidization for 5 m at room temperature. The growth conditions are same as the Sample B of the silicon substrate. Figure 4.2.3(a) is an image of the substrate after growth. It 77

89 shows that 70% of the square is covered by black-carpet-like CNTs. The black covering on the edges of the glass is the residue of silver paste used to glue the back side of the substrate to the substrate holder during the growth. Figure 4.2.3(b) is the top view SEM picture of the as-grown CNTs. It shows that the average diameter of the nanotubes is about 50 nm. Figure is SEM image of the cross section view of the sample. It is estimated that the density of the nanotube is about 2 x 10 9 /cm 2. The length of the nanotube is about 3-4 µm. The vertical alignment of the nanotube is clear to see. 78

90 a b Figure 4.3 (a) and (b) SEM images of the as-grown CNTs from different catalyst and growth conditions with the same growth temperature of 500 o C. 79

91 Figure 4.4 HRTEM image of the same sample as Fig 4.2.1(b). Figure 4.5 SEM view of the cross section of the CNT film grown on glass substrate. 80

92 a b Figure 4.6 (a) Digital camera image of the as-grown CNTs on glass substrate. (b) SEM images of the as-grown CNTs on glass substrate. The average diameter of the nanotubes is about 50 nm. 81

93 4.3 Results and Discussion Fe catalyst film without annealing in air won t lead to the growth of CNTs, when the growth temperature is lower than 700 o C with other growth parameters set to optimized values. The success of the growth of CNTs at 500 o C on the catalyst after the pretreatment is attributable to several aspects. The first aspect is the formation of the nm size catalyst islands. According to the growth mechanism described in Chapter 2, a smooth surface won t help in formation of the tubular structure of the CNTs. Annealing of the catalyst film in vacuum or in the non-oxygen (H 2, N 2, NH 3 or so) gas environment at a temperature above 700 o C can produce nm-size catalyst islands. But 700 o C is far higher than the transition temperature of soda-lime glass, which is 550 o C. So the growth temperature as well as the pretreatment temperature must be lower than 550 o C. The pretreatment temperature we chose was 500 o C. When the smooth Fe film experiences the 500 o C annealing in air for 3 h, there are mainly two transitions occurring. The first one is the oxidation of Fe to Fe 2 O 3 and the second one is the formation of Fe 2 O 3 grains on the surface. Figure 4.3.1(a) shows the SEM picture of the surface of the catalyst film after the annealing, including the trenches and protrusions. Most of the grains are smaller than 50 nm and some of them are as small as 10 nm. The morphology of the Fe 2 O 3 film depends on the annealing temperature and Fe film thickness. Figure 4.3.1(b) is the same catalyst film calcined at 700 o C for 10 m. It is evident that the aperture between each grain is larger and size of the grain is more uniform. Figure is the schematic figure of the formation of the Fe 2 O 3 grains. The introduction of the Fe 2 O 3 grains leads to the roughness of the surface, which is preserved during the reduction of Fe 2 O 3 to Fe. Therefore, by annealing of the catalyst film in air at 500 o C instead of higher temperature annealing in a non-oxygen environment, the rough surface is achieved. The 82

94 importance of the island shape of the catalyst has already been addressed in Chapter 2. Additionally, the roughness is sufficient to increase both the specific surface area and the porosity of the Fe film. Consequently, this facilitates carbon atoms reaching the catalyst. This is the one of the key reasons for success in growing CNTs at 500 o C. Figures 4.3.1(a) and (b) also show that the Fe 2 O 3 grains are small and uniform. These two characteristics lead to the small diameter and high density CNT growth. The characterization of the as-grown CNTs will be addressed later. 83

95 Figure 4.7 (a) SEM picture of the surface of the Fe catalyst after the annealing in air at 500 o C for 3 h. Figure 4.7 (b) SEM picture of the surface of the Fe catalyst after the annealing in air at 700 o C for 10 m. 84

96 Figure 4.7 (c) SEM picture of the surface of the Fe catalyst after the annealing in vacuum at 700 o C for 1 h. Fe films Substrate Annealing Fe 2 O 3 grains Substrate Figure 4.8 Schematic figures of the formation of Fe 2 O 3 on the surface of the substrate when the Fe film is annealed in air for either 550 o C or 700 o C. 85

97 a b Figure 4.9 (a) SEM image of the growth result for the substrate annealed in air. (b) SEM image of the growth result for the substrate annealed in vacuum. The growth conditions for (a) and (b) are same. Figure 4.10 Crystal structure of Fe 2 O 3. Big balls and small balls represent Fe 3+ and O 2- ions respectively. (Figure is from Mineralogy Database website.) 86

98 For the purpose of comparison, the same catalyst was also annealed at 700 o C for 1 h in a vacuum furnace. Figure 4.3.1(c) is the SEM picture of the catalyst morphology after the annealing. One big difference between the vacuum annealing and the air annealing is the superior uniformity of the catalyst calcined in air, which aids growth of uniform diameter of CNTs. The substrates annealed at 500 o C in air and 700 o C in vacuum were sent to the growth chamber for growth of CNTs under the exact same growth conditions. Figure 4.3.3(a) and (b) are the SEM pictures of the growth results of the substrates annealed in air and in vacuum, respectively. The differences between the two substrates are evident. No CNT can be found on the substrate annealed in vacuum. Even in Figure 4.3.1(c) there are many catalyst islands formed, but the formation won t help in growing CNTs at low temperature. So the formation of rough catalyst islands is not enough for the low temperature growth of CNTs. There must be other reasons for the growth. As mentioned above, the second transition is the oxidation of Fe to Fe 2 O 3. As the 5 nm thermal deposited Fe contacts the air at room temperature (at relative humidity below the critical value), it forms an invisible film of iron oxide with a thickness of a fraction of nm. This iron oxide film acts as a barrier layer between the iron and oxygen species, and the oxidation reaction rate soon falls [58]. As temperature rises, the oxidation becomes significant. A complex oxide film forms because the different oxides are stable at different oxygen partial pressures and both the oxygen concentration and equilibrium potential vary across the film. The sequence of the composition from O 2 boundary to iron core is Fe 2 O 3, Fe 3 O 4 and FeO. For the 5 nm Fe film calcined at 500 o C for 3 h, most of the grain is Fe 2 O 3. The color of the film turns from black to red, which also shows the existence of the Fe 2 O 3. The component of Fe 2 O 3 helps to grow CNTs at 500 o C in two aspects. First, during the 87

99 growth stage, the plasma of C 2 H 2 and NH 3 further reduces Fe 2 O 3 to Fe, which is the final catalyst of CNT growth. When the chamber temperature is low, the iron surface won t liquefy. The dissociated carbon atoms won t dissolve into iron islands which results in forming a layer of amorphous carbon on the surface of iron islands. The formation of amorphous carbon on the surface of the catalyst islands is regarded as a barrier layer between the iron surface and carbon atoms and high traveling speed ions inside of plasma sheath. All carbon atoms will only form layers of amorphous carbon on iron islands instead of carbon nanotubes so all catalyst is poisoned. The reduced oxygen from Fe 2 O 3 will react with those amorphous carbon atoms to form carbon monoxide and carbon dioxide and fly away. So the amorphous carbon layers will be etched and the bare surface of iron will be exposed. Under the bombardments of high speed irons, the surface of the catalyst is liquefied and carbon atoms will dissolve into catalyst islands and saturate them. Carbon nanotubes will be formed when carbon atoms precipitate from the saturated iron islands. Recent experimental results show that to effectively remove the amorphous carbon, an original solution is to add weak oxidizer to the reaction chamber. Researchers in Japan have successfully produced SWNTs with long length and high density by adding small amounts of H 2 O to the reaction chamber [37]. Research at Stanford University found that small amounts of O 2 are also helpful in growing long-length small-diameter CNTs [46]. Growing long CNTs by adding the weak oxidizer to remove the amorphous carbon on the surface of catalyst seems highly effective. The same rationale is also applied in the growth of CNTs at 500 o C, except that the oxygen atoms are from the reduction reaction of Fe 2 O 3 to Fe instead of being deliberately added. 88

100 The second reason is the barrier function of Fe 2 O 3. When there is no barrier layer, such as SiO 2, between the catalyst film and silicon substrate, the metal silicide is easily formed [40]. Part of the catalyzing power will be lost due the poisoning of the catalyst. The formation of Fe 2 O 3 calcined at low temperature is a barrier layer between the catalyst film and silicon surface. The Fe 2 O 3 layer blocks the diffusion of the silicon to iron catalyst to form iron silicide. The activity of the catalyst is preserved so that the low temperature growth of the CNTs is realized. 4.4 Conclusion The oxidizing of Fe catalyst film into Fe 2 O 3 by either air annealing at 500 o C for 3 h or O 2 plasma for 5 m was found important to realize the growth of CNTs at 500 o C. The rough, porous and uniform catalyst film created more contacting surface between the carbon and Fe species. The release of O atoms from Fe 2 O 3 could etch away amorphous carbon which would poison the catalyst so that to help the growth of CNTs at low temperature. The growth of dense (~10 9 /cm 2 ), vertical aligned and large area (~0.7 cm 2 ) of carbon nanotubes on sodalime glass substrate was achieved. 89

101 Chapter 5. Electron Field Emission of Carbon Nanotube Cathode In this chapter, the concept of field emission will be described. Consequently, the relationship between field emission and shape of emitter will be explained. Other behaviors of nanotubes in an electric field will be discussed. A simple and effective way to activate more CNT emitters will be shown. Finally, the pattern growth of CNTs on glass substrates and the field emission properties will be presented. 5.1 Introduction of Electron Field Emission Electron source technologies have had a profound impact on a whole range of applications, from flat-panel displays to electron microscopes and telecommunications. Electron emission can be achieved by the photoelectric effect, thermionic electron emission or electron field emission. The most commonly employed electron source is the thermionic cathode. In a thermionic electron source, the filament (usually made of tungsten) is heated to between 1000 o C and 3000 o C so that electrons inside of metal will gain sufficient kinetic energy to overcome the energy barrier (work function Φ, a few electron V for metals) and emit to vacuum phase. Electron field emission is fundamentally different from thermionic electron emission. Field emission is defined as the emission of electrons from the surface of a condensed phase into another phase, usually a vacuum, under the action of high electrostatic field [59]. The high electric field decreases the thickness of the potential barrier

102 to a few nm, so that the Fermi-level electrons can quantum mechanically tunnel through it and be emitted into the vacuum at ambient temperature. Figure is the schematic figure of the field emission process. Compared to thermionic electron emission, field emission offers several attractive characteristics, such as instantaneous response to applied electric field, resistance to temperature fluctuation and radiation, high degree of focus ability in electron optics, and good on/off ratio [60]. Figure 5.1 Schematic figure of the field emission process. 91

103 Fowler-Nordheim (F-N) Equation and Field Enhancement Factor β The basic physics of field emission is well understood. Based upon improved experimental results and early theoretical work, Fowler and Nordheim developed a straightforward theory in 1928 based on the modern quantum mechanics to explain the phenomenon of electron field emission [61]. The analysis of density of the emission current was shown in detail in [62-64]. According to F-N theory, for 1-D case with plane electrode configuration, the emission current density, J, can be expresses as: J 3 / 2 I B 2 Cϕ v( y) = = E0 exp[ ] A ϕt y) E ( 0 where I is emission current; A is effective emission area; φ is the work function of emitter [ev]; E 0 = V/d is the macro electrical field between cathode and anode; V is inter electrode voltage; d is the distance between electrodes; t(y) and v(y) are elliptical dimensionless Nordheim functions; B and C are constants; with B 3 q e 8πh =1.54x10-6 AeVV -2 ; C 8π (2m 3q h e e ) 1/ 2 = 6.83x10 9 ev -3/2 Vm -1 ; mass of free electron m e = 9.11x10-31 kg; elementary charge q e = 1.6x10-19 C; Planck constant, h = 6.62x10-34 Js = 4.13x10-15 evs. The values of the elliptical functions t(y) and v(y) are computed in [65, 66]. The argument of the functions t(y); v(y) is the parameter of the Nordheim [61]: 1/ 2 3 q E E e y = = πε 0ϕ ϕ where φ in ev; permittivity of free space ε 0 = 8.85x10-12 C/Vm. The value of y are belongs to the interval 0-1. Since, t(y) is a weak function of y, then it is approximated as: t 2 (y) =

104 [67]. For the real field strength at the tip of the emitter ~ (2-5)10 9 V/m the following approximation of v(y) is used usually [67]: v( y) y 2 = Taking into account of and and putting the values of B and C, could be transformed as: J 3 / 2 I B Cϕ = = exp[ ] E0 exp[ ] A.1ϕ ϕ E 1 0 For a cathode in the typical form of a flat metallic surface, according to the FN equation, the threshold field to extract an appreciable amount of electrons is around 10 4 V/µm which is impractically high. Since the local field (E l ) can be enhanced when the shape of the emitter changes, the local field can be significantly larger than the applied parallel-plane (E 0 ) field. The best shape of the emitter is found to be whisker-like [68]. Figure shows that a whisker-like emitter is better than several other types, which include sharpened pyramid shape, hemi-spheriodal shape and pyramid shape. Figure shows the electric field distribution around the whisker-like emitter. Figure 5.2 Various shapes of field emitters. (a) Rounded whisker; (b) Sharpened pyramid; (c) Hemi-spheroidal; (d) Pyramidal. (Figure is from [68].) 93

105 Figure 5.3 Electric field distribution around the tip and body of a whisker-like emitter. (Figure is from [68].) Since the voltage V is the parameter directly measured in experiments, it is more practical to define a geometrical field factor γ as: E l = γ V Taking into account of 5.1.5, equation for a single emitter can be rewritten as: I = A exp[ ] γ V ϕ ϕ ϕ exp[ γv 3 / 2 ] Apply the natural logarithm, we can obtain: ϕ 2 2 I 3 / 1 γ 10. ln( ) ln( ) A = V γ V ϕ ϕ A ln(i/v 2 ) vs. V -1 plot (F-N plot) will generate a straight line if the emission mechanism follow the F-N theory. The value γ can be calculated as: ϕ 1 γ = m b 94

106 where b is the slope of the F-N plot. The field enhancement factor is defined as β =, taking into account of 5.1.5: β = γ x d where d is the distance between cathode and anode. E l E Field Emission from Single Emitter and Multi-emitters Carbon nanotubes can be used as field emission electron sources in single and multiple electron beam devices. One possible application of a single electron beam instrument is electron microscopy that uses a single nanotube as a field emission electron gun to produce a highly coherent electron beam [69]. Electrochemically etched wires (e.g., W [70] and Au [71]) are usually used as the support material. One way to attach the nanotube to the tip of the wire tip is to use microscopy to manipulate the wire to approach and touch the tube. An individual nanotube sticks to the wire tip either by van der Walls force alone or after first applying a bit of conductive carbon glue to the tip. Another way is to use the dielectrophoresis method to affix individual nanotubes to the tip of the supporting wire [72]. Figure is the schematic drawing of the dielectrophoresis method. The polarized nanotubes (small bar in the figure) in solution were first aligned along the electric field direction due to the torque acting on the induced dipole. When the permittivity of the object is larger than that of the medium, the result shows a positive dielectrophoresis force driving the nanotubes towards the high field region. Figure is the SEM image of the final tungsten tip attached with carbon nanotube. The direct growth, also called bottom up growth of one nanotube on a support by CVD, is another method to realize the single nanotube emitter structure [73]. Field emission measurements were carried out. Field 95

107 emission energy distribution (FEED) was used to calculate the work function (Φ) and field enhancement factor (β). The work function values of 7.3 ± 0.7eV [70] ; 5.3 ev [74] ; 5.1 ev [75] and 0.5 ev ~ 2 ev [22] were reported by different groups. The reported values of field enhancement factor (β) were also in a wide range. The different methods of preparing the samples and environments may affect the measurements, which leads to the diversity of the results. However, despite the various results of field emission measurements, it shows that single nanotubes own excellent field emission properties. Figure 5.4 Schematic drawing of the dielectrophoresis method. Small bar represents carbon nanotube in the solution. Sharp needle represents the wire tip. The small figure shows the local electric field around the nanotube. (from [72]) Figure 5.5 SEM image of the final tungsten tip attached with carbon nanotube[72]. 96

108 The most famous example of the multiple electron source device is the Spindt-type cathode invented in late 1960s [76]. As illustrated in Figure (a) and (b), a Spindt-type emitter has cone geometry with a sub-micron apex radius. Typically, high-melting-point metals such as molybdenum or tungsten are used to fabricate field emitter tips. In recent years, silicon tips have also been used in field emitter arrays owing to the widespread availability of silicon microfabrication techniques. The Spindt-type cathode has been successfully used in prototypes for flat panel display and microwave amplifiers. But the working vacuum required by the Spindt-type cathode made of metals is quite high (> 10-9 Torr) because ion bombardment from residual gas species will blunt the emitter cones and reduce the lifetime of the device. Moreover, this kind of cathode suffers from high manufacturing cost, which also limits wide application. Therefore, alternative materials to replace metal tips in the Spindt-type cathode structure have been actively pursued. Metal or Silicon Tips 1-2 µm Metal Gate Film SiO 2 Dielectric 1-3 µm Si Base Figure 5.6 (a) Schematic drawing of the cross section view of a Spindt-type cathode. 97

109 Figure 5.6 (b) Scanning electron microfigure of a single gated nanocone structure [77]. Methods to Pattern Carbon Nanotube Film onto Cathode Substrate One important step in using carbon nanotubes as electron emitters in microelectronic devices is to apply them in patterns onto the substrates. This can be realized either by producing the nanotubes (by arc discharge, laser ablation or CVD) and subsequently patterning them onto the substrates ( top-down process) or by growing the nanotubes directly on a substrate pre-patterned with catalyst ( bottom-up process). Screen-printing and electrophoresis are the typical examples of top-down process and all kinds of chemical vapor deposition are examples of bottom-up process. For more information, please refer to Section Behavior of Carbon Nanotubes in Electric Field Understanding the behavior, besides field emission, of carbon nanotubes in an electric field can help us design better CNT-based field emission cathodes and look for techniques to improve the field emission properties of cathodes. It is important to study the carbon nanotube s behaviors in vacuum since electron field emissions happen in a vacuum 98

110 environment. Additionally, some popular methods to assemble nanotube to substrate, such as EPD, require a liquid phase, which makes it equally important to study them in liquid phase. In Liquid Phase As mentioned above, dielectrophoresis has been used to assemble the CNT fibrils at the tip of the AFM cantilever. Dielectrophoresis is defined as the lateral motion imparted on uncharged particles as a result of polarization induced by non-uniform electric fields [78]. The nanotubes produced by arc discharge or laser ablation were purified and cut into several µm-long pieces in strong acid [72] and dispersed into de-ionized water with aid of sonication. The experimental setup comprises of two electrodes: a conductive sharp needle and a flat metal counter electrode, which are inserted into the nanotube-water solution and used to produce the non-uniform electric field. An alternating current (AC) field is applied between the two electrodes. Because the carbon nanotube is highly anisotropic, the polarizability tensors are consequently highly anisotropic, with the result that the dipole moment (µ z ) along the nanotube cylindrical axis directions is larger than the that (µ a ) perpendicular to the cylindrical axis directions [79]. Therefore, the transitional forces (F = µ E) along the nanotube cylindrical axis directions are dominated so that nanotubes align their cylindrical axis with the field directions. Furthermore, the polarity of this force depends on the polarity of the induced dipole moment, which in turn is determined by the conductivity and permittivities of nanotube and its suspending medium, de-ionized water in this case [80]. Nanotube is more polarisable than the de-ionized water and is attracted towards the strong field at the pin electrode. When the needle is drawn out from the solution, nanotubes continuously migrate and attach to form a fibril. So the actions of nanotubes in water can be expressed as a series of steps, alignment-migration-assembly-continuous growth [81]. 99

111 Figure illustrates these steps. Additionally, since the polarity of the induced force is related to the conductivity and permittivity of nanotube, which are not related to the size and shape of the tubes, dielectrophoresis can be used to separate metallic nanotubes from semiconducting nanotubes [82]. Electrophoresis is another widely used method to assemble the carbon nanotube onto substrates. Small amounts of charger are required and added to the carbon nanotube solution. The charger ions attach to the nanotube body, and both the ions and nanotubes migrate to the electrode under the electric field and attach onto it. The direction that the nanotubes move is determined by the polarity of the charger and direction of the electric field. It is unlike dielectrophoresis, where the moving direction is independent of the polarity of electrodes. So direct current (DC) and uniform electric field are often used in the electrophoresis method [83]. Figure shows an example of the phenomenon of electrophoresis. 100

112 Figure 5.7 Proposed mechanism by which the individual CNTs assemble into uniform fibrils under an asymmetrical electric field. (Figure is from [81].) 101

113 a b Figure 5.8 (a) SWNT-assembly and orientation in response to an applied DC electric field. (b) High magnification image of lower part of the deposition. (Figures are from [84].) In Vacuum Phase Due to the small size of carbon nanotubes, the direct measurements of mechanical and electrical properties of individual nanotubes or nanotube films under an electric field are done under AFM, SEM and TEM. When a static electric field is applied between a MWNT and the counter electrode, the carbon nanotube can be charged by this applied electric field. The induced charge is distributed mostly at the tip of nanotube. The electrostatic force between the charge at the tip of nanotube and counter electrode results in the deflection of the nanotube [85]. Figure shows the electrostatic attraction between two carbon nanotubes induced by a constant field across the electrodes. The nanotube is very flexible and can bend by almost 90 o. 102

114 Figure 5.9 Electrostatic attraction between two carbon nanotubes induced by a constant field across the electrodes. The induced charges are mainly accumulated at the tip. (Figure is from [85].) When a time-dependent electric field is applied, the dynamic deflections occur in accordance with the applied electric field. By changing the frequency of the applied potential, the first and second order harmonic oscillations are observed. Figure (a-c) shows this phenomenon. 103

115 Figure 5.10 Nanotube response to resonant alternating applied potentials. (a) In the absence of a potential, the nanotube tip vibrated slightly because of thermal effect. (b) The first order harmonic oscillation. (c) The second order harmonic oscillation. (Figure is from [86].) The in situ TEM method is also used to study the electron field emission properties of several nanotubes. Figure shows the image of the electron field emissions. The dark contrast near the tips of the nanotube is the field contributed by the charges on the tip of the nanotube and the emitting electrons. By examining the intensity and the area of the darkness of the tips, the intensity of the electron emission from different nanotube tips can be compared. The TEM image indicates that the electron emissions are not synchronized among all tubes. The tubes standing out from the surrounding tube have more intense electron emission due to the higher enhanced local electric field around the tips as indicated by white arrows in the image. Due to the interference by higher neighbor tubes, the lower tubes 104

116 experience a weaker local enhanced field, so that the emission is not that intense as indicated by black arrows. Figure 5.11 In situ TEM observation of the electric-field-induced electron emission from carbon nanotubes. The applied voltage is 60 V, and the emission current ~20µA. (Figure is from [85].) The deformation of nanotubes in accordance with the applied electric field can be permanent if electron field emission is involved. A SEM is used to study the geometry 105

117 change of a nanotube that is used as a field emitter [87]. When a relatively weak bias potential is applied between the nanotube and counter electrode, the attraction force will bend the nanotube toward the counter electrode. Once the bias potential is removed, the nanotube, being flexible, will return to the original position with little deviation. When the applied electric field is strong enough, electrons will be extracted from the nanotube. Depending on the amount of the emission current and period of emission, the deformation can be permanent regardless of the existence or removal of the external applied electric field. Figure shows the comparison of the shapes of the same nanotube before and after the field emission. In Figure (a) there was a little curve close to the top of the nanotube. After the field emission, the tube became straight, pointing to the counter electrode without the previous little curve as seen from Figure (b). So the deformation remained after the removal of the applied electric field. During field emission, the structure of the nanotube can be impaired, which can affect the field emission performance and lifetime of the CNT-based cathode. Figure is the SEM image depicting the shortening of the nanotube after field emission. After a period of field emission, another SEM image was taken and compared to the SEM image taken before the emission [87]. The comparison showed that the length of the nanotube was shortened by about 10%. This shortening was caused mainly by bombardment by ions. Even when the pressure inside the SEM chamber was about 10-8 Torr [87], there was still a small amount of residual gases left in the chamber. During the electron emission, the high-speed electrons could ionize those gases to produce exciters. The ionized ions would follow the direction of the electric field to the tip of the emission nanotube and bombard the surface of the nanotube. 106

118 The bombardment can be severe enough to sputter the carbon atoms out so that the length of the nanotube is shortened. 107

119 Figure 5.12 (a) The shape of the nanotube before field emission; (b) The shape of the same nanotube after the field emission. This SEM image was taken after the electric field was removed. The nanotube stayed in the straight shape. The inset to the right-hand side of each image is an enlarged look at the nanotube. (Figures are from [87].) Figure 5.13 CNT was shortened after field emission. (a) Before field emission; (b) After field emission; (c) Side-by-side comparison of the CNT before and after field emission. The nanotube was about 10% shorter. (Figures are from [87].) 108

120 5.4 Techniques to Improve the Field Emission Properties of Carbon Nanotube Cathodes For CNT film emitters, there are many techniques to improve field emission performance. For example the soft rubber roller technique applies to the screen printing method [88]; adhesive taping [89] is applicable to the screen printing method [90], EPD method [54] and CVD method. Additional methods include DC current treatment [91], multiple cycles [88] and laser irradiation [92]. Whatever the post treatment method, the essential point of post treatment is to achieve the desired CNT morphology for better field emission properties. The desired morphologies include vertical alignment, desired density and similar height of CNTs. The advantage of MPECVD, as mentioned before, is its ability to fabricate vertically aligned carbon nanotubes. Moreover, the height of CNTs in the growth area is uniform. The deposition pattern can be realized by photolithography. Here we will report an effective and simple method utilizing a cotton swab to activate more CNT emitters Experiment Using photolithography, the pattern shown in Figure of Fe film (5 nm thick) was deposited on a Si wafer. Each square was 24 x 24 µm; the pitch was 50 µm. The entire area covered by Fe film was 1 cm x 1 cm. After depositing Fe film, the substrate was transferred to the growth chamber. The chamber was first pumped down to 1 x 10-3 Torr. Then a gas mixture of 10 sccm H 2 / 190 sccm N 2 was introduced to increase the pressure to 27 Torr. The temperature of the chamber was increased to 820 o C by heating filament. Then 150 sccm NH 3 was introduced to replace H 2 /N 2 gas. A 350 W microwave was transferred to the growth chamber to ignite the NH 3 plasma for 3 m. Immediately, 30 sccm C 2 H 2 gas was introduced to 109

121 the growth chamber. The growth time was 1.5 m. After the growth was done, the sample was examined by optical microscope and SEM. Figure (a) is the 45 o tilt angle view SEM of the pattern and (b) is the SEM image of one square of the CNT pattern. 50 µm Fe Si 24 µm Figure 5.14 Schematic drawing of the details of the pattern. 110

122 Figure 5.15 (a) 45 o angle view SEM of the CNT pattern. Figure 5.15 (b) High magnification SEM image of one of the CNT squares. 111

123 5.4.2 Field Emission and the effect of Cotton Swab Scratch At first, the substrate was put into the vacuum chamber to measure the field emission without any treatment. Then the substrate was taken out and some scratches were made on specific areas. Immediately the substrate was put into the vacuum chamber and a second field emission reading was taken. It was obvious that there were more emitters in the scratched areas and that the density of emitters was higher than in the unscratched area. Therefore, the second incidence of scratch was made on the rest of dark area of the cathode. Then field emission was measured for the third time. This time all areas of the cathode emitted and this emission was uniform. Figure (a) is the digital camera image of the first instance of field emission, (b) is the image of the second instance and (c) is the image of the third instance. These three images indicate that more emitters appeared on the scratched location and that finally the emission covered the entire cathode and was uniform. Figure shows the I-V curves of those three instances of field emission measurements. The left shift of the curves means the turn-on voltage decreased and emission current density increased at the same electric field. 112

124 Figure 5.16 (a) Digital camera image of the first instance of field emission. Figure 5.16 (b) Digital camera image of the second instance of field emission. Figure 5.16 (c) Digital camera image of the third instance of field emission. 113

125 8.0x x10-3 I-V curves of the three tests 1st FE 2ndFE 3rd FE 6.0x x10-3 I (A) 4.0x x x x x E (V/um) Figure 5.17 The I-V curves of those three instances of field emission measurements Results After the scratches, the original pattern was changed. Figure is the optical microscope image of the pattern after scratches. It shows that some CNTs in the squares were displaced from their original places, as pointed out by the red square in Figure The actual size of the CNT square was shrunk. The square s being smaller reduced the screening effect, so that the local electric field on the tips of CNTs was larger than before. That is the first reason that there are more emitters after the scratches. The second reason is that some CNT clusters were brought to the places that were free of CNTs, as indicated by the red circle in Figure The new appearance of CNT clusters resulted in new emitters, which 114

126 emitted when the electric field was applied. So after the scratches, there were more emitters appearing. Figure 5.18 Microscope image of the area after scratches. The red square shows the shrinkage of the emission site. The red circle shows the appearance of new emitter site. 115

127 5.5 Growth Pattern of Carbon Nanotubes on Glass Substrate and Field Emission Properties The successful synthesis of MWNTs on glass substrate at 500 o C and the growth mechanism were described in Chapter 4. In this section the field emission of the CNT film, as well as patterning of the CNT film, will be described Introduction The importance to fabricate CNTs on glass substrate has been illustrated in Chapter 1 and Chapter 4. The synthesis of CNTs or carbon nanofibres (CNFs) on glass substrate by PECVD can be found in [57] and [30]. But the crystallization of the CNTs and CNFs are poor. The ballistic electron transferring of CNT is based on the calculation of perfect crystallization in a nanotube model [93]. When CNTs or CNFs are not well crystallized, the increased scattering between the electrons and phonons will impair their field emission properties. Therefore, the fabrication of well-crystallized CNTs on glass substrate at low temperature presents a challenge Synthesis and Characterization A 2.5 cm (or 3 cm) by 2.5 cm soda-lime glass slide was used as the substrate. Ultrasonic baths in acetone and IPA for 3 m each and ultraviolet ozone treatment for 3 m were used to clean the surface of the substrate. Due to the high thermal stability, low electric resistance and good barrier characteristics, titanium nitride (TiN) layer with 100 nm thick was sputtered onto the surface of the substrate by PVD as the underlayer of Fe catalyst film. An aluminum metal plate with a 1 cm by 1 cm window and No. 200 metal meshes (SPI ) were used as shadow masks to make the pattern. 5 nm thick iron (Fe) film was deposited onto the TiN layer by thermal evaporation. After the deposition of the patterned Fe film, the substrate was 116

128 transferred to growth chamber in air. The chamber was pumped down to 1x10-3 Torr first and then a gas mixture of 10 sccm H 2 / 190 sccm N 2 was introduced to increase the pressure to 30 Torr. At the same time, the heating filament heated the chamber and the substrate to 500 o C. After 10 m of stabilizing, 150 sccm NH 3 was introduced to replace the H 2 /N 2. A microwave with power of 500 W was transferred to the growth chamber to ignite NH 3 plasma. After 3 m of NH 3 plasma treatment of the substrate, the pressure was increased to 42 Torr and microwave power was increased to 550 W. Another 3 m of NH 3 plasma treatment was added. Finally, 30 sccm C 2 H 2 gas was introduced to the chamber for 1 m to deposit CNT film. Figure shows the digital camera image of the substrate with the CNT pattern. The square area is the Fe film. The black covering is the CNT film. The image shows that 80% of the entire pattern area is covered by CNTs. Figure (a) is an SEM image of the top view of the CNT film. The image shows that the density of the as-grown CNTs is very high. Figure (b) is an SEM image of a place scratched by tweezers. The vertically aligned nanotubes are lying down on the substrate. This picture indicates that the length of as-grown CNTs is about 3 µm. The straight body of the CNTs suggests good crystallization. Figure (a) is a 45 o tilted angle, low magnification SEM image of the patterned CNT film. The roughness on the substrate is the deformation of the glass substrate due to the plasma bombardment. Figure (b) is the higher magnification 90 o tilt angle SEM image of the CNT film, which illustrates that the nanotubes are vertically aligned and well patterned. The density of the as-grown CNT is roughly 2x10 9 /cm 2 within the Fe covering area. Figure (a) shows a hollow nanotube with diameter of 10 nm. Figure (b) shows a bamboo structure nanotube with diameter of 20 nm. Those two structures are typical of MWNT. The 117

129 good crystallization illustrated by HRTEM images will be responsible for reduction of local heating due to the field emission and longer lifetime of field emission. Figure 5.19 Digital camera image of the substrate with the CNT pattern. Figure 5.20 (a) Top view SEM image of as-grown CNT film. 118

130 Figure 5.20 (b) SEM image of the area scratched by tweezers. The length of the as-grown nanotube is about 3~4 µm. The crystallization is good. Figure 5.21 (a) SEM 45 o angle tilted angle view of the CNT pattern. 119

131 Figure 5.21 (b) 90 o angle SEM image of a cross section with higher magnification. Figure 5.22 (a) HRTEM image of a hollow nanotube with diameter of 11 nm. 120

132 Figure 5.22 (b) HRTEM image of a bamboo structure nanotube with diameter of 20 nm Field Emission Measurements and Results Field emission measurements were taken in the vacuum chamber with the pressure of ~5x10-7 Torr. A Keithley Model 248 High Voltage Supply was used to provide high voltage signal. A Keithley 2000 Multimeter was used to measure the current in the circuit. A DEI PVX-4100 Pulse Generator was used to tune the signal into pulse signal to protect the anode from continuous electron bombardments. A 2.5cm by 2.5 cm home made phosphor screen was used as an anode to collect electrons as well as to show the emission site density and emission uniformity. A pair of 200 µm thick glass slides was used as spacers between the anode and the cathode. The I-V curve was measured with a 10% duty cycle. Both the 121

133 patterned CNT film and unpatterned CNT film were measured. Figure is the figure to show the comparison of the two curves. The figure clearly shows that the turn-on field decreased from 8.6 V/µm to 2.4 V/µm. The voltages to achieve current density of 2.0 ma/cm 2 were 13 V/µm and 4.5V/µm for unpatterned and patterned CNT films, respectively. Digital camera images of the light emission from the phosphor screen were also taken for comparison. Figure (a) and (b) are the digital camera images of unpatterned and patterned CNT films, respectively. The emission site density increased, though not as remarkably as the I-V curve improved. 2.5x10-3 patterned unpattern 2.0x x10-3 I (A) 1.0x x E (V/um) Figure 5.23 Solid-line curve is the I-V curve of the patterned CNT film. The hollow circle dotted-line curve is the I-V curve of the unpatterned CNT film and the hollow triangle dotted-line curve is I-V curve of the patterned CNT film. 122

134 1cm Figure 5.24 (a) Digital camera image of the emission of unpatterned CNT film. 0.5cm Figure 5.24 (b) Digital camera image of the field emission of the patterned CNT film An Empirical Model to Calculate Field Enhancement Factor Due to the screening effect, field emission β from CNT film is always lower than β of an individual CNT emitter. An empirical equation is achieved by Bonard [94]: β = β [1 exp( 2.32s / )] film 0 h 123